Self-Efficacy and Performance of Research Skills Among First-Semester Bioscience Doctoral Students
This is an R Markdown Notebook detailing all of the data analysis completed for the paper entitled “Self-Efficacy and Performance of Research Skills Among First-Semester Bioscience Doctoral Students”.
All analyses were performed by K. Lachance. This notebook was compiled on May 22, 2020.
Library Import
Import libraries and packages required for analysis and visualization and set seed for reproducibility.
# Install packages, if not already installed
list.of.packages <- c("reshape2", "ggplot2", "gridExtra", "grid", "orddom", "svglite", "ggdendro", "plotly")
new.packages <- list.of.packages[!(list.of.packages %in% installed.packages()[,"Package"])]
if(length(new.packages)) install.packages(new.packages)
# Load packages
library(reshape2) # Data structure manipulation
library(ggplot2) # Plotting
library(gridExtra) # Plotting
library(grid) # Plotting
library(orddom) # Statistics## Loading required package: psych##
## Attaching package: 'psych'## The following objects are masked from 'package:ggplot2':
##
## %+%, alphalibrary(svglite) # Plotting
library(ggdendro) # Plotting dendrograms
library(plotly) # Plotting dendrograms##
## Attaching package: 'plotly'## The following object is masked from 'package:ggplot2':
##
## last_plot## The following object is masked from 'package:stats':
##
## filter## The following object is masked from 'package:graphics':
##
## layout# Make jitter plots and other analyses / visualizations reproducible
set.seed(123)Data Import
Import data, contained in ExperimentalDesignData.csv file. Each row corresponds to a student (identified with an anonymous code) and each column contains information about that student’s response about research skills self-efficacy, experimental design aptitude, experience, and demographic information. In total, 103 students were surveyed (only students that completed all pre- and post-test assessments were included in this study).
# Main data
data = read.csv(file = "./StudentData.csv", header = TRUE, row.names = 1, stringsAsFactors = FALSE) # Import data
numStudents = nrow(data) # Calculate number of students in cohort to be analyzed
# Print output
cat(paste("There are ", numStudents, " student responses contained in the following analyses.\n", sep ="")) ## There are 103 student responses contained in the following analyses.# Factors in experience and comfort
data2 = read.csv(file = "./ExperienceComfortFactors.csv", header = TRUE, stringsAsFactors = FALSE) # Import data for Fig 4
# Demographics by year for Table S1
data3 = read.csv(file = "./DemographicsByYear.csv", header = TRUE, row.names = 1, stringsAsFactors = FALSE) # Import data for Table S1
# Demographics by sex for Table S4
data4 = read.csv(file = "./DemographicsBySex.csv", header = TRUE, row.names = 1, stringsAsFactors = FALSE) # Import data for Table S4Figure 1
Length of time spent doing pre-doctoral research is not predictive of student research skills self-efficacy or self-reported experience and comfort with experimental design.
Figure 1A
# Make a data frame holding frequency of students' previous research experience
labExp <- melt(table(data$LabExp))
colnames(labExp) = c("Years", "Freq")
labExp$Years = as.character(labExp$Years)
labExp$Percent = round((labExp$Freq / numStudents) * 100, 0)
# Make pie chart
labExp$Years <- factor(labExp$Years, levels = c("7", "6", "5", "4", "3", "2", "1")) # Re-order for plotting
fig1a <- ggplot(labExp, aes(x="", y=Percent, fill=Years))+
geom_bar(width = 1, stat = "identity") +
coord_polar("y", start=0) +
scale_fill_manual("Previous Lab Experience", values=c("#333333", "#4C4C4C", "#666666", "#808080", "#999999", "#B2B2B2", "#CCCCCC"),
labels=c(paste("> 7 years (", labExp$Percent[7], "%)", sep=""),
paste("6 - 7 years (", labExp$Percent[6], "%)", sep=""),
paste("5 - 6 years (", labExp$Percent[5], "%)", sep=""),
paste("4 - 5 years (", labExp$Percent[4], "%)", sep=""),
paste("3 - 4 years (", labExp$Percent[3], "%)", sep=""),
paste("2 - 3 years (", labExp$Percent[2], "%)", sep=""),
paste("< 2 years (", labExp$Percent[1], "%)", sep="")))+
theme_minimal() +
theme(axis.title = element_blank(), panel.border = element_blank(), panel.grid=element_blank(), axis.ticks = element_blank(), axis.text=element_blank()) +
ggtitle("Figure 1A") + theme(plot.title = element_text(hjust = 0.5))
ggsave(filename = "./Figures/Fig1A.svg", plot = fig1a, width = 6, height = 4)
plot(fig1a)
Figure 1B
# Split data by responses to self-efficacy questionaire before and after the first semester of graduate school
SE_pre = data[,14:27]
SE_post = data[,28:41]
SE_pre$LabExp = as.character(data$LabExp)
SE_post$LabExp = as.character(data$LabExp)
# Melt responses
mSE_pre = melt(SE_pre)## Using LabExp as id variablesmSE_post = melt(SE_post)## Using LabExp as id variables# Add question number label to each pre / post array
mSE_pre$Question = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = numStudents)
mSE_pre$Test = "Pre"
mSE_post$Question = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = numStudents)
mSE_post$Test = "Post"
# Change labels to strings for categorical plotting
mSE_pre[mSE_pre$value==1,"value"] = "a1" # Not at all, pre-test
mSE_pre[mSE_pre$value==2,"value"] = "b1" # A little, pre-test
mSE_pre[mSE_pre$value==3,"value"] = "c1" # A moderate amount, pre-test
mSE_pre[mSE_pre$value==4,"value"] = "d1" # A lot, pre-test
mSE_pre[mSE_pre$value==5,"value"] = "e1" # A great deal, pre-test
mSE_post[mSE_post$value==1,"value"] = "a2" # Not at all, post-test
mSE_post[mSE_post$value==2,"value"] = "b2" # A little, post-test
mSE_post[mSE_post$value==3,"value"] = "c2" # A moderate amount, post-test
mSE_post[mSE_post$value==4,"value"] = "d2" # A lot, post-test
mSE_post[mSE_post$value==5,"value"] = "e2" # A great deal, post-test
# Stacked bar chart
labexp.labs = c("< 2 years", "2 - 3 years", "3 - 4 years", "4 - 5 years", "5 - 6 years", "6 - 7 years", "> 7 years")
names(labexp.labs) = as.character(1:7)
fig1b <- ggplot(mSE_pre, aes(Test)) +
geom_bar(aes(fill=value), position = "fill") + facet_grid(~ LabExp, labeller = labeller(LabExp = labexp.labs)) +
theme_bw() +
scale_fill_manual("Research Skills Self-Efficacy", values = c("a1" = "#DAE3F3", "b1" = "#B4C7E7", "c1" = "#8FAADC", "d1" = "#2F5597", "e1" = "#203864"), labels = c("Not at all", "A little", "A moderate amount", "A lot", "A great deal")) +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
theme(axis.title=element_blank(), axis.text=element_blank(), axis.ticks=element_blank(), strip.background = element_blank()) +
ggtitle("Figure 1B") + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig1b, filename="./Figures/Fig1B.svg", width = 4, height = 3)
plot(fig1b)
Figure 1Ci
# Calculate linear regression
lm_exp = lm(formula = ExperimetalDeignExperience ~ LabExp, data = data)
summary(lm_exp) # Adjusted R-squared: 0.06328##
## Call:
## lm(formula = ExperimetalDeignExperience ~ LabExp, data = data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.29102 -0.79102 -0.08613 0.70898 1.91387
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 2.67634 0.24372 10.981 < 2e-16 ***
## LabExp 0.20490 0.07294 2.809 0.00597 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.013 on 101 degrees of freedom
## Multiple R-squared: 0.07247, Adjusted R-squared: 0.06328
## F-statistic: 7.891 on 1 and 101 DF, p-value: 0.005966# Scatter plot
fig1ci <- ggplot(data, aes(x = LabExp, y = ExperimetalDeignExperience)) +
geom_jitter(width = 0.25, height = 0.25, size = 2, color = "#808080") +
scale_x_continuous("Previous years working in lab", breaks=c(1:7), limits=c(0.5,7.5), labels = c("< 2 years", "2 - 3 years", "3 - 4 years", "4 - 5 years", "5 - 6 years", "6 - 7 years", "> 7 years")) +
scale_y_continuous("Self-reported experience\nwith experimental design", breaks=c(1:5), limits=c(0.5,5.5), labels = c("None", "Minimal", "Some", "Moderate", "Extensive")) +
theme_bw() +
theme(panel.grid.minor = element_blank()) +
ggtitle(expression(paste("Figure 1Ci (", R^2," = ", 0.06, ")", sep=""))) + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig1ci, filename="./Figures/Fig1Ci.svg", width = 4, height = 3)
plot(fig1ci)
Figure 1Cii
# Histogram
fig1cii <- ggplot(data, aes(x = ExperimetalDeignExperience)) +
geom_histogram(binwidth = 0.5, fill = "#174E94") +
theme_bw() +
scale_x_continuous("Self-reported experience\nwith experimental design", breaks=c(1:5), labels = c("None", "Minimal", "Some", "Moderate", "Extensive")) +
ylim(0, 40) + ylab("Number of students") +
coord_flip() +
theme(panel.grid.minor = element_blank(), panel.grid.major = element_blank()) +
ggtitle("Figure 1Cii") + theme(plot.title = element_text(hjust = 0.5))
plot(fig1cii)
ggsave(fig1cii, filename="./Figures/Fig1Cii.svg", width = 4, height = 3)Figure 1Di
# Calculate linear regression
lm_comf = lm(formula = ExperimetalDeignComfort ~ LabExp, data = data)
summary(lm_comf) # Adjusted R-squared: 0.05171 ##
## Call:
## lm(formula = ExperimetalDeignComfort ~ LabExp, data = data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3.736 -1.148 0.381 1.264 2.852
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.5593 0.3838 9.274 3.57e-15 ***
## LabExp 0.2942 0.1149 2.562 0.0119 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.595 on 101 degrees of freedom
## Multiple R-squared: 0.06101, Adjusted R-squared: 0.05171
## F-statistic: 6.562 on 1 and 101 DF, p-value: 0.01189# Scatter plot
fig1di <- ggplot(data, aes(x = LabExp, y = ExperimetalDeignComfort)) +
geom_jitter(width = 0.25, height = 0.25, size = 2, color = "#808080") +
scale_x_continuous("Previous years working in lab", breaks=c(1:7), limits=c(0.5,7.5), labels = c("< 2 years", "2 - 3 years", "3 - 4 years", "4 - 5 years", "5 - 6 years", "6 - 7 years", "> 7 years")) +
scale_y_continuous("Self-reported comfort\nwith experimental design", breaks=c(1:7), limits=c(0.5,7.5), labels = c("Extremely uncomfortable", "Moderately uncomfortable", "Slightly uncomfortable", "Neither comfortable nor uncomfortable", "Slightly comfortable", "Moderately comfortable", "Extremely comfortable")) +
theme_bw() +
theme(panel.grid.minor = element_blank()) +
ggtitle(expression(paste("Figure 1Di (", R^2," = ", 0.05, ")", sep=""))) + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig1di, filename="./Figures/Fig1Di.svg", width = 4, height = 3)
plot(fig1di)
Figure 1Dii
# Histogram
fig1dii <- ggplot(data, aes(x = ExperimetalDeignComfort)) +
geom_histogram(binwidth = 0.5, fill = "#174E94") +
theme_bw() +
scale_x_continuous("Self-reported Comfort\nwith experimental design", breaks=c(1:7), labels = c("Extremely uncomfortable", "Moderately uncomfortable", "Slightly uncomfortable", "Neither comfortable nor uncomfortable", "Slightly comfortable", "Moderately comfortable", "Extremely comfortable")) +
ylim(0, 40) + ylab("Number of students") +
coord_flip() +
theme(panel.grid.minor = element_blank(), panel.grid.major = element_blank()) +
ggtitle("Figure 1Dii") + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig1dii, filename="./Figures/Fig1Dii.svg", width = 4, height = 3)
plot(fig1dii)
Figure 2
Students improved their performance of experimental design over the first semester of doctoral training.
Figure 2A
# Note that only 2017 data can be used in BEDCI calculations
data_2017 = data[data$Year == "2017",]
numStudents_2017 = nrow(data_2017) # Calculate number of students in cohort to be analyzed (2017)
# Print output
cat(paste("There are ", numStudents_2017, " student responses contained in the following BEDCI analyses.\n", sep ="")) ## There are 45 student responses contained in the following BEDCI analyses.# Calculate linear regression
lm_bedci = lm(formula = Pre_CITotal ~ LabExp, data = data_2017)
summary(lm_bedci) # Adjusted R-squared: 0.018##
## Call:
## lm(formula = Pre_CITotal ~ LabExp, data = data_2017)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3.9738 -1.5019 0.2621 1.2621 3.0262
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 8.2660 0.6137 13.470 <2e-16 ***
## LabExp 0.2359 0.1755 1.344 0.186
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.878 on 43 degrees of freedom
## Multiple R-squared: 0.04032, Adjusted R-squared: 0.018
## F-statistic: 1.806 on 1 and 43 DF, p-value: 0.186# Plot histograms
fig2a <- ggplot(data_2017, aes(x = (Pre_CITotal / 14) * 100, y = LabExp)) +
geom_jitter(width = 0.25, height = 0.25, size = 2, color = "#174E94") +
coord_flip() +
scale_x_continuous("Total BEDCI score (%)", breaks=seq(40, 100, 20), limits=c(30, 90)) +
scale_y_continuous("Previous years working in lab", breaks=c(1:7), limits=c(0.5,7.5), labels = c("< 2 years", "2 - 3 years", "3 - 4 years", "4 - 5 years", "5 - 6 years", "6 - 7 years", "> 7 years")) +
theme_bw()+
theme(panel.grid.minor = element_blank()) +
ggtitle(expression(paste("Figure 2A (", R^2," = ", 0.02, ")", sep=""))) + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig2a, filename="./Figures/Fig2A.svg", width = 4, height = 3)
plot(fig2a)
Figure 2B
# Calculate scores (in percent)
pre_score = data.frame(Score = (data_2017$Pre_CITotal / 14) * 100, Test = "Pre", stringsAsFactors = FALSE)
post_score = data.frame(Score = (data_2017$Post_CITotal / 14) * 100, Test = "Post", stringsAsFactors = FALSE)
BEDCI_score = rbind(pre_score, post_score)
# Perform wilcoxon signed-rank test
wTest = wilcox.test(data_2017$Pre_CITotal, data_2017$Post_CITotal, paired=TRUE, alternative = "two.sided", exact = FALSE)
# Determine significance level
sig1 = "is NOT a statistically significant difference"
sig2 = ""
if (wTest$p.value < 0.001) {
sig1 = "IS a statistically significant difference"
sig2 = "***"
} else if (wTest$p.value < 0.01) {
sig1 = "IS a statistically significant difference"
sig2 = "**"
} else if (wTest$p.value < 0.05) {
sig1 = "IS a statistically significant difference"
sig2 = "*"
}
# Print output
cat(paste("By a Wilcoxon signed-rank test, there ", sig1, " between student performance on the BEDCI concept inventory before and after their first semester of graduate school (V = ", wTest$statistic, ", p = ", round(wTest$p.value, 4), sig2, ")", sep=""))## By a Wilcoxon signed-rank test, there IS a statistically significant difference between student performance on the BEDCI concept inventory before and after their first semester of graduate school (V = 202, p = 0.0079**)# Order for plotting
BEDCI_score$Test <- factor(BEDCI_score$Test, levels = c('Pre','Post'), ordered = TRUE)
# Create violin plot
fig2b <- ggplot(BEDCI_score, aes(x = Test, y = Score)) +
geom_violin(draw_quantiles = c(0.5), scale = "width", aes(fill = Test)) +
geom_jitter(height = 3, width = 0.1) +
scale_fill_manual(values = c("#174E94", "#EE7D31")) +
scale_y_continuous("Total BEDCI score (%)", breaks = seq(40, 100, 20), labels = seq(40, 100, 20)) +
scale_x_discrete(labels = c("Pre-test", "Post-test")) +
theme_bw() +
theme(panel.grid.minor = element_blank(), panel.grid.major = element_blank(), axis.title.x = element_blank(), legend.position="none") +
ggtitle("Figure 2B") + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig2b, filename="./Figures/Fig2B.svg", width = 2, height = 3)
plot(fig2b)
Figure 2C
# Function for calculating standard error of the mean
stderr_perc <- function(x, na.rm=FALSE) {
if (na.rm) x <- na.omit(x)
y = (x / numStudents_2017) * 100
sqrt(var(y))
}
# Make data frame suitable for plotting
Q = c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14") # Order of questions by subject matter
mCI_pre = data.frame(Question = Q, Test = "Pre", Score = (colSums(data_2017[,42:55]) / numStudents_2017) * 100, SEM = apply(as.matrix(data_2017[,42:55]), 2, stderr_perc), stringsAsFactors = FALSE) # Calculate mean score for each question on the pre-test
mCI_post = data.frame(Question = Q, Test = "Post", Score = (colSums(data_2017[,57:70]) / numStudents_2017) * 100, SEM = apply(as.matrix(data_2017[,42:55]), 2, stderr_perc), stringsAsFactors = FALSE) # Calculate mean score for each question on the post-test
# Combine
mCI = rbind(mCI_pre, mCI_post)
# Function to perform McNemar's chi-squared test for each question on the BEDCI pre- and post-tests
mcNemarChiSquaredTest <- function(dat, Q) {
# Get relevant data for each question
pre = dat[,(Q+41)]
post = dat[,(Q+56)]
# Calculate values to populate matrix
c_c = sum(pre + post == 2) # Correct, correct
i_c = sum(pre - post == -1) # Incorrect, correct
c_i = sum(pre - post == 1) # Correct, incorrect
i_i = sum(pre + post == 0) # Incorrect, incorrect
# Create confusion matrix
mat = matrix(c(c_c, i_c, c_i, i_i), nrow = 2, dimnames = list("Pre" = c("Correct", "Incorrect"), "Post" = c("Correct", "Incorrect")))
# Calculate significance
mcnemar.test(mat)
}
# Calculate the p-value for each question and save to an output table
coreConcepts = c("Controls", "", "Hypotheses", "", "Biological variation", "", "Accuracy", "Extraneous factors", "", "Independent sampling", "", "Random sampling", "", "Purpose of experiments")
questions = c(1, 5, 2, 9, 3, 10, 4, 6, 14, 7, 12, 8, 13, 11) # Questions ordered by subject
sigOut = data.frame(CoreConcept = coreConcepts, Question = questions, Chi = rep(0, 14), pValue = rep(0, 14), Significance = rep("", 14), stringsAsFactors = FALSE)
i = 1 # Iterate over every row of output
for (q in questions) {
mnTest = mcNemarChiSquaredTest(data_2017, q)
sigOut$Chi[i] = round(mnTest$statistic, 2)
sigOut$pValue[i] = round(mnTest$p.value, 4)
i = i + 1
}
pValues = p.adjust(sigOut$pValue, method = "fdr")
for (i in 1:14) {
sigOut$pAdj[i] = round(pValues[i], 4)
if (sigOut$pAdj[i] < 0.001) {
sigOut$Significance[i] = "***"
} else if (sigOut$pAdj[i] < 0.01) {
sigOut$Significance[i] = "**"
} else if (sigOut$pAdj[i] < 0.05) {
sigOut$Significance[i] = "*"
}
}
# Print signficance table
print(sigOut) # Also Table S3## CoreConcept Question Chi pValue Significance pAdj
## 1 Controls 1 4.00 0.0455 0.1592
## 2 5 0.10 0.7518 1.0000
## 3 Hypotheses 2 5.88 0.0153 0.1592
## 4 9 0.00 1.0000 1.0000
## 5 Biological variation 3 0.00 1.0000 1.0000
## 6 10 4.00 0.0455 0.1592
## 7 Accuracy 4 4.92 0.0265 0.1592
## 8 Extraneous factors 6 0.12 0.7237 1.0000
## 9 14 0.12 0.7237 1.0000
## 10 Independent sampling 7 0.00 1.0000 1.0000
## 11 12 0.75 0.3865 1.0000
## 12 Random sampling 8 0.00 1.0000 1.0000
## 13 13 0.00 1.0000 1.0000
## 14 Purpose of experiments 11 0.08 0.7728 1.0000# Reorder for plotting
mCI$Question <- factor(mCI$Question, levels = c('Q1', 'Q5', 'Q2', 'Q9', 'Q3', 'Q10', 'Q4', 'Q6', 'Q14', 'Q7', 'Q12', 'Q8', 'Q13', 'Q11'), ordered = TRUE)
mCI$Test <- factor(mCI$Test, levels = c('Pre','Post'), ordered = TRUE)
# Create bar chart
fig2c <- ggplot(mCI, aes(x = Question, y = Score, fill = Test, group = Test)) +
geom_bar(position = position_dodge(), stat = "identity") +
geom_errorbar(aes(ymin=Score-SEM, ymax=Score+SEM), colour="black", width = 0.3, position = position_dodge(1)) +
scale_fill_manual("", values = c("#174E94", "#EE7D31"), labels = c("Pre-test", "Post-test")) +
ylab("% of students providing expert-like answers") +
theme_bw() +
theme(panel.grid.minor = element_blank(), panel.grid.major.x = element_blank(), axis.title.x = element_blank()) +
ggtitle("Figure 2C") + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig2c, filename="./Figures/Figure2C.svg", width = 5, height = 3)
plot(fig2c)
Figure 3
Students significantly improved in many aspects of research skills self-efficacy during their first semester of doctoral training.
Figure 3B
# Test for significance
wilcoxonSignedRankTest <- function(dat, Q) {
# Get the data of interest for each question
pre = dat[,Q+13]
post = dat[,(Q+27)]
# Perform wilcoxon signed-rank test for significance
wilcox.test(pre, post, paired=TRUE, alternative = "two.sided", exact=FALSE)
}
# Calculate the p-value for each question and save to an output table
Items = c("Understand contemporary concepts in your field", "Make use of the primary scientific research literature in your field (e.g., journal articles)", "Identify a specific question for investigation based on the research in your field", "Formulate a research hypothesis based on a specific question", "Design an experiment or theoretical test of the hypothesis", "Understand the importance of 'controls' in research", "Observe and collect data", "Statistically analyze data", "Interpret data by relating results to the original hypothesis", "Reformulate your original research hypothesis (as appropriate)", "Relate your results to the 'bigger picture' in your field", "Orally communicate the results of research projects", "Write a research paper for publication", "Think independently")
sigOut = data.frame(Item = Items, Question = 1:14, V = rep(0, 14), pValue = rep(0, 14), Significance = rep("", 14), stringsAsFactors = FALSE)
for (i in 1:14) {
wTest = wilcoxonSignedRankTest(data, i)
sigOut$V[i] = wTest$statistic
sigOut$pValue[i] = round(wTest$p.value, 4)
}
wTest_Total = wilcox.test(melt(SE_pre)$value, melt(SE_post)$value, paired=TRUE, alternative = "two.sided", exact=FALSE)## Using LabExp as id variables
## Using LabExp as id variablestmp = data.frame(Item = "Total", Question = 15, V = wTest_Total$statistic, pValue = round(wTest_Total$p.value, 4),Significance = "", stringsAsFactors = F)
sigOut[15,] = tmp
pValues = p.adjust(sigOut$pValue, method = "fdr")
for (i in 1:15) {
sigOut$pAdj[i] = round(pValues[i], 4)
if (sigOut$pAdj[i] < 0.001) {
sigOut$Significance[i] = "***"
} else if (sigOut$pAdj[i] < 0.01) {
sigOut$Significance[i] = "**"
} else if (sigOut$pAdj[i] < 0.05) {
sigOut$Significance[i] = "*"
}
}
# Print signficance table
print(sigOut) # This is also Table S2## Item
## 1 Understand contemporary concepts in your field
## 2 Make use of the primary scientific research literature in your field (e.g., journal articles)
## 3 Identify a specific question for investigation based on the research in your field
## 4 Formulate a research hypothesis based on a specific question
## 5 Design an experiment or theoretical test of the hypothesis
## 6 Understand the importance of 'controls' in research
## 7 Observe and collect data
## 8 Statistically analyze data
## 9 Interpret data by relating results to the original hypothesis
## 10 Reformulate your original research hypothesis (as appropriate)
## 11 Relate your results to the 'bigger picture' in your field
## 12 Orally communicate the results of research projects
## 13 Write a research paper for publication
## 14 Think independently
## V Total
## Question V pValue Significance pAdj
## 1 1 524.5 0.0071 * 0.0118
## 2 2 488.0 0.0784 0.0905
## 3 3 677.0 0.0164 * 0.0224
## 4 4 429.5 0.0000 *** 0.0000
## 5 5 339.5 0.0000 *** 0.0000
## 6 6 389.0 0.0033 ** 0.0062
## 7 7 607.5 0.3072 0.3072
## 8 8 498.0 0.0004 ** 0.0012
## 9 9 385.5 0.0028 ** 0.0060
## 10 10 286.0 0.0000 *** 0.0000
## 11 11 506.0 0.1198 0.1284
## 12 12 413.0 0.0024 ** 0.0060
## 13 13 411.0 0.0208 * 0.0260
## 14 14 619.0 0.0114 * 0.0171
## V 15 90521.0 0.0000 *** 0.0000# Combine melted self-efficacy data frames
mSE = rbind(mSE_pre, mSE_post)
# Re-order for plotting
mSE$Test = factor(mSE$Test, levels = c("Pre", "Post"))
mSE$Question = factor(mSE$Question, levels = c("Q1", "Q2", "Q3", "Q4", "Q5", "Q14", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13"))
mSE$value = factor(mSE$value, levels = c("a1", "b1", "c1", "d1", "e1", "a2", "b2", "c2", "d2", "e2"))
# Stacked bar chart
fig3bi <- ggplot(mSE, aes(Test)) +
geom_bar(aes(fill=value), position = "fill") + facet_grid(~ Question) +
theme_bw() +
scale_fill_manual("Research Skills\nSelf-Efficacy",
values = c("a1" = "#DAE3F3", "a2" = "#FBE5D6", "b1" = "#B4C7E7", "b2" = "#F8CBAD", "c1" = "#8FAADC", "c2" = "#F4B183", "d1" = "#2F5597", "d2" = "#C55A11", "e1" = "#203864", "e2" = "#843C0C"), labels = c("", "", "", "", "", "Not at all", "A little", "A moderate amount", "A lot", "A great deal")) +
guides(fill=guide_legend(ncol = 2)) +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
theme(axis.title=element_blank(), axis.text=element_blank(), axis.ticks=element_blank(), strip.background = element_blank()) +
ggtitle("Figure 3Bi") + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig3bi, filename="./Figures/Figure3Bi.svg", width = 5, height = 3)
plot(fig3bi)
Figure 3Bii
# Get total self-efficacy calculation
mSE_total = data.frame(Test = rep(c("Pre", "Post"), each = 5), Lab = row.names(table(mSE$value)), Freq = as.vector(table(mSE$value)), stringsAsFactors = FALSE)
mSE_total$Test <- factor(mSE_total$Test, levels = c("Pre", "Post"), ordered = TRUE)
mSE_total$Lab = factor(mSE_total$Lab, levels = c("a1", "b1", "c1", "d1", "e1", "a2", "b2", "c2", "d2", "e2"))
# Stacked bar chart
fig3bii <- ggplot(mSE_total, aes(x=Test, y=Freq)) +
geom_bar(aes(fill=Lab), stat = "identity", position = "fill") +
theme_bw() +
scale_fill_manual("Research Skills\nSelf-Efficacy",
values = c("a1" = "#DAE3F3", "a2" = "#FBE5D6", "b1" = "#B4C7E7", "b2" = "#F8CBAD", "c1" = "#8FAADC", "c2" = "#F4B183", "d1" = "#2F5597", "d2" = "#C55A11", "e1" = "#203864", "e2" = "#843C0C"), labels = c("", "", "", "", "", "Not at all", "A little", "A moderate amount", "A lot", "A great deal")) +
guides(fill=guide_legend(ncol = 2)) +
xlab("Total") +
theme(panel.grid.minor = element_blank(), panel.grid.major = element_blank(), axis.text=element_blank(), axis.title.y=element_blank(), axis.ticks = element_blank()) +
ggtitle("Figure 3Bii") + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig3bii, filename="./Figures/Figure3Bii.svg", width = 1, height = 3)
plot(fig3bii)
Figure 3C
SE_pre$Sex = as.character(data$Sex)
SE_post$Sex = as.character(data$Sex)
# Remove NA Sex
SE_pre = SE_pre[!is.na(SE_pre$Sex),]
SE_post = SE_post[!is.na(SE_post$Sex),]
# Remove lab experience columns
SE_pre = SE_pre[,-15]
SE_post = SE_post[,-15]
male_SE_pre = SE_pre[SE_pre$Sex == "Male",]
female_SE_pre = SE_pre[SE_pre$Sex == "Female",]
male_SE_post = SE_post[SE_post$Sex == "Male",]
female_SE_post = SE_post[SE_post$Sex == "Female",]
# Melt responses
male_mSE_pre = melt(male_SE_pre)## Using Sex as id variablesmale_mSE_post = melt(male_SE_post)## Using Sex as id variablesfemale_mSE_pre = melt(female_SE_pre)## Using Sex as id variablesfemale_mSE_post = melt(female_SE_post)## Using Sex as id variablesnMale = nrow(male_SE_pre)
nFemale = nrow(female_SE_pre)
# Add question number label to each pre / post array
male_mSE_pre$Question = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = nMale)
male_mSE_pre$Test = "Pre"
male_mSE_post$Question = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = nMale)
male_mSE_post$Test = "Post"
female_mSE_pre$Question = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = nFemale)
female_mSE_pre$Test = "Pre"
female_mSE_post$Question = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = nFemale)
female_mSE_post$Test = "Post"
# Change labels to strings for categorical plotting
male_mSE_pre[male_mSE_pre$value==1,"value"] = "m_a1" # Not at all, pre-test
male_mSE_pre[male_mSE_pre$value==2,"value"] = "m_b1" # A little, pre-test
male_mSE_pre[male_mSE_pre$value==3,"value"] = "m_c1" # A moderate amount, pre-test
male_mSE_pre[male_mSE_pre$value==4,"value"] = "m_d1" # A lot, pre-test
male_mSE_pre[male_mSE_pre$value==5,"value"] = "m_e1" # A great deal, pre-test
male_mSE_post[male_mSE_post$value==1,"value"] = "m_a2" # Not at all, post-test
male_mSE_post[male_mSE_post$value==2,"value"] = "m_b2" # A little, post-test
male_mSE_post[male_mSE_post$value==3,"value"] = "m_c2" # A moderate amount, post-test
male_mSE_post[male_mSE_post$value==4,"value"] = "m_d2" # A lot, post-test
male_mSE_post[male_mSE_post$value==5,"value"] = "m_e2" # A great deal, post-test
# Change labels to strings for categorical plotting
female_mSE_pre[female_mSE_pre$value==1,"value"] = "f_a1" # Not at all, pre-test
female_mSE_pre[female_mSE_pre$value==2,"value"] = "f_b1" # A little, pre-test
female_mSE_pre[female_mSE_pre$value==3,"value"] = "f_c1" # A moderate amount, pre-test
female_mSE_pre[female_mSE_pre$value==4,"value"] = "f_d1" # A lot, pre-test
female_mSE_pre[female_mSE_pre$value==5,"value"] = "f_e1" # A great deal, pre-test
female_mSE_post[female_mSE_post$value==1,"value"] = "f_a2" # Not at all, post-test
female_mSE_post[female_mSE_post$value==2,"value"] = "f_b2" # A little, post-test
female_mSE_post[female_mSE_post$value==3,"value"] = "f_c2" # A moderate amount, post-test
female_mSE_post[female_mSE_post$value==4,"value"] = "f_d2" # A lot, post-test
female_mSE_post[female_mSE_post$value==5,"value"] = "f_e2" # A great deal, post-test
# Combine pre- and post-test dataframes for plotting
mSE <- rbind(male_mSE_pre, male_mSE_post, female_mSE_pre, female_mSE_post)
head(mSE)mSE_total = data.frame(Sex = rep(c("Female", "Male"), each = 10), Test = rep(c("Pre", "Post"), 5), Lab = row.names(table(mSE$value)), Freq = as.vector(table(mSE$value)), stringsAsFactors = FALSE)
mSE_total$Test <- factor(mSE_total$Test, levels = c("Pre", "Post"), ordered = TRUE)
# Stacked bar chart
tests.labs = c("Pre-test", "Post-test")
names(tests.labs) = c("Pre", "Post")
# Stacked bar chart
fig3c <- ggplot(mSE_total, aes(x=Sex, y=Freq)) +
geom_bar(aes(fill=Lab), stat = "identity", position = "fill") +
theme_bw() +
facet_grid(~ Test, labeller = labeller(Test = tests.labs)) +
scale_fill_manual("legend", values = c("f_a1" = "#DAE3F3", "f_b1" = "#B4C7E7", "f_c1" = "#8FAADC", "f_d1" = "#2F5597", "f_e1" = "#203864", "f_a2" = "#FBE5D6", "f_b2" = "#F8CBAD", "f_c2" = "#F4B183", "f_d2" = "#C55A11", "f_e2" = "#843C0C", "m_a1" = "#DAE3F3", "m_b1" = "#B4C7E7", "m_c1" = "#8FAADC", "m_d1" = "#2F5597", "m_e1" = "#203864", "m_a2" = "#FBE5D6", "m_b2" = "#F8CBAD", "m_c2" = "#F4B183", "m_d2" = "#C55A11", "m_e2" = "#843C0C")) +
scale_x_discrete(labels = c("Women", "Men", "Women", "Men")) +
theme(panel.grid.minor = element_blank(), panel.grid.major = element_blank(), axis.title=element_blank(), axis.text.y=element_blank(), axis.ticks = element_blank()) +
theme(strip.background = element_blank(), panel.border = element_rect(colour = "white"), legend.position = "none") +
ggtitle("Figure 3C") + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig3c, filename="./Figures/Figure3C.svg", width = 4, height = 3)
plot(fig3c)
Figure 4
Students with a net increase in research skills self-efficacy select different factors as the most important for contributing to their experience or comfort with experimental design.
Figure 4A
data2$ExpPercent = round((data2$ExperienceFreq / sum(data2$ExperienceFreq)) * 100, 2)
data2$Category <- factor(data2$Category, levels = c("Sem", "CC", "Discussing", "Present", "Lab", "Read", "Mentor", "Write"), ordered = TRUE)
fig4a <- ggplot(data2, aes(x="", y=ExpPercent, fill=Category))+
geom_bar(width = 1, stat = "identity") +
coord_polar("y", start=0) +
scale_fill_manual("Most important factor", values = c("Sem" = "#BEBEBE", "CC" = "#D64051", "Discussing" = "#F68E5C", "Present" = "#FDE08C", "Lab" = "#FBF8C0", "Read" = "#E4EA9A", "Mentor" = "#99D094", "Write" = "#3089BE"), labels = c(paste("Attending research seminars (", round(data2$ExpPercent[1], 0), "%)", sep=""),
paste("Completing coursework (", round(data2$ExpPercent[2], 0), "%)", sep=""),
paste("Discussing scientific topics with colleagues (", round(data2$ExpPercent[3], 0), "%)", sep=""),
paste("Giving scientific presentations (", round(data2$ExpPercent[4], 0), "%)", sep=""),
paste("Participating in laboratory research (", round(data2$ExpPercent[5], 0), "%)", sep=""),
paste("Reading scientific literature (", round(data2$ExpPercent[6], 0), "%)", sep=""),
paste("Recieving advice from mentors (", round(data2$ExpPercent[7], 0), "%)", sep=""),
paste("Writing a project proposal (", round(data2$ExpPercent[8], 0), "%)", sep=""))) +
theme_minimal() +
theme(axis.title = element_blank(), panel.border = element_blank(), panel.grid=element_blank(), axis.ticks = element_blank(), axis.text=element_blank()) +
ggtitle("Figure 4A") + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig4a, filename="./Figures/Figure4A.svg", width = 3, height = 3)## Warning: Removed 10 rows containing missing values (position_stack).plot(fig4a)## Warning: Removed 10 rows containing missing values (position_stack).
Figure 4B
data2$ComfPercent = round((data2$ComfortFreq / sum(data2$ComfortFreq)) * 100, 2)
fig4b <- ggplot(data2, aes(x="", y=ComfPercent, fill=Category))+
geom_bar(width = 1, stat = "identity") +
coord_polar("y", start=0) +
scale_fill_manual("Most important factor", values = c("Sem" = "#BEBEBE", "CC" = "#D64051", "Discussing" = "#F68E5C", "Present" = "#FDE08C", "Lab" = "#FBF8C0", "Read" = "#E4EA9A", "Mentor" = "#99D094", "Write" = "#3089BE"), labels = c(paste("Attending research seminars (", round(data2$ComfPercent[1], 0), "%)", sep=""),
paste("Completing coursework (", round(data2$ComfPercent[2], 0), "%)", sep=""),
paste("Discussing scientific topics with colleagues (", round(data2$ComfPercent[3], 0), "%)", sep=""),
paste("Giving scientific presentations (", round(data2$ComfPercent[4], 0), "%)", sep=""),
paste("Participating in laboratory research (", round(data2$ComfPercent[5], 0), "%)", sep=""),
paste("Reading scientific literature (", round(data2$ComfPercent[6], 0), "%)", sep=""),
paste("Recieving advice from mentors (", round(data2$ComfPercent[7], 0), "%)", sep=""),
paste("Writing a project proposal (", round(data2$ComfPercent[8], 0), "%)", sep=""))) +
theme_minimal() +
theme(axis.title = element_blank(), panel.border = element_blank(), panel.grid=element_blank(), axis.ticks = element_blank(), axis.text=element_blank()) +
ggtitle("Figure 4B") + theme(plot.title = element_text(hjust = 0.5))
ggsave(fig4b, filename="./Figures/Figure4B.svg", width = 3, height = 3)## Warning: Removed 10 rows containing missing values (position_stack).plot(fig4b)## Warning: Removed 10 rows containing missing values (position_stack).
Figure S1
65% of students had a net increase in performance of experimental design over the semester.
Figure S1A
# Calculate total change in BEDCI score (post - pre)
CI_delta = data.frame(Score = (data_2017$Post_CITotal - data_2017$Pre_CITotal), Gender = data_2017$Sex, Exp = data_2017$LabExp, stringsAsFactors = FALSE)
# Add color labels
CI_delta$Col = "orange"
CI_delta[CI_delta$Score < 0, 'Col'] ="red"
CI_delta[CI_delta$Score > 0, 'Col'] = "yellow"
cat(paste(round(sum((CI_delta$Score > 0) / numStudents_2017) * 100, 0), "% of students had a net increase in performance of experimental design over the semester.\n", sep=""))## 64% of students had a net increase in performance of experimental design over the semester.# Create histogram
figS1a = ggplot(CI_delta, aes(x = Score, fill = Col)) +
geom_histogram(binwidth = 0.5) +
scale_fill_manual("Net change in\nBEDCI score", values = c("yellow" = "#FDCB41", "orange" = "#F37627", "red" = "#811617"),
labels = c("Net decrease in concept inventory score", "No change in concept inventory score", "Net increase in concept inventory score")) +
scale_x_continuous(breaks=seq(-4, 4, 1)) +
theme_bw() +
xlab("Change in BEDCI score") + ylab("Number of students") +
theme(panel.background=element_blank(), panel.grid.minor=element_blank(), plot.background=element_blank()) +
ggtitle("Figure S1A") + theme(plot.title = element_text(hjust = 0.5))
# Plot figure
ggsave(figS1a, filename="./Figures/FigureS1A.svg", width = 6, height = 4)
plot(figS1a)
Figure S1Bi
negBEDCI_gender = data.frame(Var1 = c("Female", "Male"), Freq = c(nrow(CI_delta[CI_delta$Score < 0 & CI_delta$Gender == "Female",]), nrow(CI_delta[CI_delta$Score < 0 & CI_delta$Gender == "Male",])), stringsAsFactors = FALSE)
figS1bi <- ggplot(negBEDCI_gender, aes(x="", y=Freq, fill = Var1))+
geom_bar(width = 1, stat = "identity") +
coord_polar("y", start=0) +
scale_fill_manual("Gender of students with\nnet negative BEDCI scores", values = c("magenta4", "green4"),
labels = c(paste("Women (", round((negBEDCI_gender$Freq[1] / sum(negBEDCI_gender$Freq))*100, 0), "%)", sep=""),
paste("Men (", round((negBEDCI_gender$Freq[2] / sum(negBEDCI_gender$Freq))*100, 0), "%)", sep=""))) +
theme_minimal() +
theme(axis.title = element_blank(), panel.border = element_blank(), panel.grid=element_blank(), axis.ticks = element_blank(), axis.text=element_blank()) +
ggtitle("Figure S1Bi") + theme(plot.title = element_text(hjust = 0.5))
ggsave(figS1bi, filename="./Figures/FigureS1Bi.svg", width = 3, height = 3)
plot(figS1bi)
Figure S1Bii
negBEDCI_labExp = data.frame(Var1 = as.character(1:7), Freq = c(nrow(CI_delta[CI_delta$Score < 0 & CI_delta$Exp == 1,]), nrow(CI_delta[CI_delta$Score < 0 & CI_delta$Exp == 2,]), nrow(CI_delta[CI_delta$Score < 0 & CI_delta$Exp == 3,]), nrow(CI_delta[CI_delta$Score < 0 & CI_delta$Exp == 4,]), nrow(CI_delta[CI_delta$Score < 0 & CI_delta$Exp == 5,]), nrow(CI_delta[CI_delta$Score < 0 & CI_delta$Exp == 6,]), nrow(CI_delta[CI_delta$Score < 0 & CI_delta$Exp == 7,])), stringsAsFactors = FALSE)
negBEDCI_labExp$Var1 <- factor(negBEDCI_labExp$Var1, levels = as.character(1:7))
figS1bii <- ggplot(negBEDCI_labExp, aes(x="", y=Freq, fill = Var1))+
geom_bar(width = 1, stat = "identity") +
coord_polar("y", start=0, direction = -1) +
scale_fill_manual("Lab experience of students with\nnet negative BEDCI scores", values=c("7" = "#333333", "6" = "#4C4C4C", "5" = "#666666", "4" = "#808080", "3" = "#999999", "2" = "#B2B2B2", "1" = "#CCCCCC"),
labels = c(paste("< 2 years (", round((negBEDCI_labExp$Freq[1] / sum(negBEDCI_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("2 - 3 years (", round((negBEDCI_labExp$Freq[2] / sum(negBEDCI_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("3 - 4 years (", round((negBEDCI_labExp$Freq[3] / sum(negBEDCI_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("4 - 5 years (", round((negBEDCI_labExp$Freq[4] / sum(negBEDCI_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("5 - 6 years (", round((negBEDCI_labExp$Freq[5] / sum(negBEDCI_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("6 - 7 years (", round((negBEDCI_labExp$Freq[6] / sum(negBEDCI_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("> 7 years (", round((negBEDCI_labExp$Freq[7] / sum(negBEDCI_labExp$Freq)) * 100, 0), "%)", sep=""))) +
theme_minimal() +
theme(axis.title = element_blank(), panel.border = element_blank(), panel.grid=element_blank(), axis.ticks = element_blank(), axis.text=element_blank()) +
ggtitle("Figure S1Bii") + theme(plot.title = element_text(hjust = 0.5))
ggsave(figS1bii, filename="./Figures/FigureS1Bii.svg", width = 3, height = 3)
plot(figS1bii)
Figure S2
Correlation generated sever topical groups of research skills self-efficacy questions.
Figure S2A
# Split data by responses to self-efficacy questionaire before and after the first semester of graduate school
SE_pre = data[,14:27]
# Melt responses
mSE_pre = melt(SE_pre)## No id variables; using all as measure variables# Add question number label to each pre / post array
Questions = c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14")
mSE_pre$Question = rep(Questions, each = numStudents)
# Make data frame for clustering
out = data.frame(matrix(0, nrow = length(Questions), ncol = length(Questions)), row.names = Questions, stringsAsFactors = FALSE)
colnames(out) = Questions
i = 1
for (Qi in Questions) {
j = 1
for (Qj in Questions) {
out[i, j] = round(cor(mSE_pre[mSE_pre$Question == Qi, "value"], mSE_pre[mSE_pre$Question == Qj, "value"], method = "spearman"), 2)
j = j + 1
}
i = i + 1
}
# Cluster
row.order <- hclust(dist(out, method = "euclidean"), method="ward.D")$order
col.order <- hclust(dist(t(out), method = "euclidean"), method="ward.D")$order
out_new <- out[row.order, col.order]
m_out <- melt(as.matrix(out_new))
names(m_out)[c(1:2)] <- c("Qx", "Qy")
# Dendrogram
dd.row <- as.dendrogram(hclust(dist(t(out), method = "euclidean"), method="ward.D"))
dx <- dendro_data(dd.row)
# Helper function for creating dendograms
ggdend <- function(df) {
ggplot() +
geom_segment(data = df, aes(x=x, y=y, xend=xend, yend=yend)) +
theme_bw() +
labs(x = "", y = "") + theme_minimal() +
theme(axis.text = element_blank(), axis.ticks = element_blank(), panel.grid = element_blank())
}
# x/y dendograms
figS2a_dendrogramX = ggdend(dx$segments)
figS2a_heatmap = ggplot(data = m_out, aes(x = Qx, y = Qy, fill = 1 - value)) +
geom_raster() +
theme_bw() +
scale_fill_distiller("Spearman correlation", palette = "Blues",) + # pre = blues, post = oranges, delta = greys
theme(axis.text.x = element_text(angle = 45, hjust = 1), axis.title = element_blank(), axis.ticks =element_blank()) +
theme(panel.border = element_blank(), panel.grid.major = element_blank(), panel.grid.minor = element_blank(), legend.position = "bottom")
figS2a = grid.arrange(figS2a_dendrogramX, figS2a_heatmap, ncol = 1, heights = c(1, 3), top ="Figure S2A")
ggsave(figS2a, filename="./Figures/FigureS2A.svg", width = 6, height = 4)Figure S2B
# Split data by responses to self-efficacy questionaire before and after the first semester of graduate school
SE_post = data[,28:41]
# Melt responses
mSE_post = melt(SE_post)## No id variables; using all as measure variables# Add question number label to each pre / post array
Questions = c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14")
mSE_post$Question = rep(Questions, each = numStudents)
# Make data frame for clustering
out = data.frame(matrix(0, nrow = length(Questions), ncol = length(Questions)), row.names = Questions, stringsAsFactors = FALSE)
colnames(out) = Questions
i = 1
for (Qi in Questions) {
j = 1
for (Qj in Questions) {
out[i, j] = round(cor(mSE_post[mSE_post$Question == Qi, "value"], mSE_post[mSE_post$Question == Qj, "value"], method = "spearman"), 2)
j = j + 1
}
i = i + 1
}
# Cluster
row.order <- hclust(dist(out, method = "euclidean"), method="ward.D")$order
col.order <- hclust(dist(t(out), method = "euclidean"), method="ward.D")$order
out_new <- out[row.order, col.order]
m_out <- melt(as.matrix(out_new))
names(m_out)[c(1:2)] <- c("Qx", "Qy")
# Dendrogram
dd.row <- as.dendrogram(hclust(dist(t(out), method = "euclidean"), method="ward.D"))
dx <- dendro_data(dd.row)
# x/y dendograms
figS2b_dendrogramX = ggdend(dx$segments)
figS2b_heatmap = ggplot(data = m_out, aes(x = Qx, y = Qy, fill = 1 - value)) +
geom_raster() +
theme_bw() +
scale_fill_distiller("Spearman correlation", palette = "Oranges",) + # pre = blues, post = oranges, delta = greys
theme(axis.text.x = element_text(angle = 45, hjust = 1), axis.title = element_blank(), axis.ticks =element_blank()) +
theme(panel.border = element_blank(), panel.grid.major = element_blank(), panel.grid.minor = element_blank(), legend.position = "bottom")
figS2b = grid.arrange(figS2b_dendrogramX, figS2b_heatmap, ncol = 1, heights = c(1, 3), top ="Figure S2B")
ggsave(figS2b, filename="./Figures/FigureS2B.svg", width = 6, height = 4)Figure S2C
# Split data by responses to self-efficacy questionaire before and after the first semester of graduate school
SE_delta = data[,28:41] - data[,14:27]
# Melt responses
mSE_delta = melt(SE_delta)## No id variables; using all as measure variables# Add question number label to each pre / post array
Questions = c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14")
mSE_delta$Question = rep(Questions, each = numStudents)
# Make data frame for clustering
out = data.frame(matrix(0, nrow = length(Questions), ncol = length(Questions)), row.names = Questions, stringsAsFactors = FALSE)
colnames(out) = Questions
i = 1
for (Qi in Questions) {
j = 1
for (Qj in Questions) {
out[i, j] = round(cor(mSE_delta[mSE_delta$Question == Qi, "value"], mSE_delta[mSE_delta$Question == Qj, "value"], method = "spearman"), 2)
j = j + 1
}
i = i + 1
}
# Cluster
row.order <- hclust(dist(out, method = "euclidean"), method="ward.D")$order
col.order <- hclust(dist(t(out), method = "euclidean"), method="ward.D")$order
out_new <- out[row.order, col.order]
m_out <- melt(as.matrix(out_new))
names(m_out)[c(1:2)] <- c("Qx", "Qy")
# Dendrogram
dd.row <- as.dendrogram(hclust(dist(t(out), method = "euclidean"), method="ward.D"))
dx <- dendro_data(dd.row)
# x/y dendograms
figS2c_dendrogramX = ggdend(dx$segments)
figS2c_heatmap = ggplot(data = m_out, aes(x = Qx, y = Qy, fill = 1 - value)) +
geom_raster() +
theme_bw() +
scale_fill_distiller("Spearman correlation", palette = "Greys",) + # pre = blues, post = oranges, delta = greys
theme(axis.text.x = element_text(angle = 45, hjust = 1), axis.title = element_blank(), axis.ticks =element_blank()) +
theme(panel.border = element_blank(), panel.grid.major = element_blank(), panel.grid.minor = element_blank(), legend.position = "bottom")
figS2c = grid.arrange(figS2c_dendrogramX, figS2c_heatmap, ncol = 1, heights = c(1, 3), top ="Figure S2C")
ggsave(figS2c, filename="./Figures/FigureS2C.svg", width = 6, height = 4)Figure S3
Over 20% of students decrease on at least one individual self-efficacy item from the beginning to the end of the semester.
Figure S3A
# Make combined data frame
mSE_pre_delta = cbind(mSE_pre, mSE_delta)[,c(3,2,5)]
colnames(mSE_pre_delta) <- c("Question", "Pre", "Delta")
# Create color scheme
mSE_pre_delta$Color = "orange" # If delta = 0
mSE_pre_delta[mSE_pre_delta$Delta > 0, "Color"] = "red" # if delta > 0
mSE_pre_delta[mSE_pre_delta$Delta < 0, "Color"] = "yellow" # if delta < 0
# Plot scatterplot
figS3a <- ggplot(mSE_pre_delta, aes(Pre, Delta, color = Color)) +
geom_jitter(height = 0.25, width = 0.25, alpha = 0.4, size = 3, aes(shape = Question)) +
scale_shape_manual(values=1:14) +
scale_color_manual(values = c("yellow" = "#811617", "orange" = "#F37627", "red" = "#FDCB41"), guide = FALSE) +
geom_abline(intercept = 0.5, slope = -1, linetype="dashed", color = "grey") +
geom_abline(intercept = 5.5, slope = -1, linetype="dashed", color = "grey") +
scale_x_continuous("Self-reported self-efficacy at\nbeginning of semester", limits = c(0.5, 5.5), breaks = 1:5, labels = c("Not at all", "A little", "A moderate amount", "A lot", "A great deal")) +
scale_y_continuous("Incremental difference\nin Likert scale rating", limits = c(-4.5, 3.5), breaks = -4:3) +
theme_bw() +
guides(shape=guide_legend(ncol=2)) + labs(shape = "Self-efficacy question") +
theme(panel.grid.minor = element_blank()) +
ggtitle("Figure S3A") + theme(plot.title = element_text(hjust = 0.5))
plot(figS3a)
ggsave(figS3a, file="~/Desktop/FigS3A.svg", width = 6, height = 4)Figure S3B
# Construct a dataframe for a density plot of changes in student responses
SE_change = data.frame(red = rowSums(SE_delta < 0), orange = rowSums(SE_delta == 0), yelloq = rowSums(SE_delta > 0), stringsAsFactors = FALSE)
# Melt data frame for plotting
mSE_change = melt(SE_change)## No id variables; using all as measure variablesfigS3b <- ggplot(mSE_change, aes(x = value, fill = variable, color = variable)) +
geom_density(alpha = 0.7) +
theme_bw() +
scale_fill_manual("Net change in self-efficacy", values = c("#811617", "#F37627", "#FDCB41"), labels = c("Decreased self-efficacy", "No change", "Increased self-efficacy")) +
scale_color_manual("Net change in self-efficacy", values = c("#811617", "#F37627", "#FDCB41"), labels = c("Decreased self-efficacy", "No change", "Increased self-efficacy")) +
scale_x_continuous("Total items of self-efficacy\nchanged over time", breaks=seq(0, 14, 2), labels=seq(0, 14, 2),limits=c(0,14)) +
scale_y_continuous("Percent of students (%)", breaks=c(0, 0.1, 0.2), labels=c(0, 10, 20)) +
theme(panel.grid.minor = element_blank(), panel.grid.major= element_blank()) +
ggtitle("Figure S3B") + theme(plot.title = element_text(hjust = 0.5))
plot(figS3b)
ggsave(figS3b, file="~/Desktop/FigS3B.svg", width = 6, height = 4)Figure S4
Over 70% of students had a net increase in research skills over the semester.
# Add columns to delta- dataframe to identify each student (currently row names)
SE_delta$Student = row.names(SE_delta)
# Change question names
colnames(SE_delta) = c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14", "Student")
# Melt data
mSE_delta = melt(SE_delta)## Using Student as id variablescolnames(mSE_delta) <- c("Student", "Question", "Delta")
# Calcualte frequencies at which students improved are decreased in self-efficacy
delta_freq = dcast(mSE_delta, Student~Delta)## Using Delta as value column: use value.var to override.## Aggregation function missing: defaulting to length# Calcualte net-change for each student, adding information about gender and years of expeirence working in a lab
delta_net = rowSums(cbind(delta_freq[,2]*-4, delta_freq[,3]*-3, delta_freq[,4]*-2, delta_freq[,5]*-1, 0, delta_freq[,7], delta_freq[,8]*2, delta_freq[,9]*3))
delta_net = data.frame(Value = delta_net, Gender = data$Sex, Exp = data$LabExp, stringsAsFactors = F)
# Add color labels
delta_net$Color = "orange"
delta_net[delta_net$Value < 0, 'Color'] ="red"
delta_net[delta_net$Value > 0, 'Color'] = "yellow"
# Create histogram
figS4a = ggplot(delta_net, aes(x = Value, fill = Color)) +
geom_histogram(binwidth = 0.5) +
scale_fill_manual("Net change in\nself-efficacy", values = c("yellow" = "#FDCB41", "orange" = "#F37627", "red" = "#811617"),
labels = c("Net decrease in self-efficacy score", "No change in self-efficacy score", "Net increase in self-efficacy score")) +
scale_x_continuous("Net change in self-efficacy", breaks=seq(-20, 20, 5)) +
scale_y_continuous("Number of students", breaks=seq(0, 10, 2)) +
theme_bw() +
theme(panel.background=element_blank(), panel.grid.minor=element_blank(), plot.background=element_blank()) +
ggtitle("Figure S4A") + theme(plot.title = element_text(hjust = 0.5))
ggsave(figS4a, filename="./Figures/FigureS4A.svg", width = 6, height = 4)
plot(figS4a)
Figure S4Bi
negSE_gender = data.frame(Var1 = c("Female", "Male"), Freq = c(nrow(delta_net[delta_net$Value < 0 & delta_net$Gender == "Female",]), nrow(delta_net[delta_net$Value < 0 & delta_net$Gender == "Male",])), stringsAsFactors = FALSE)
figS4bi <- ggplot(negSE_gender, aes(x="", y=Freq, fill = Var1))+
geom_bar(width = 1, stat = "identity") +
coord_polar("y", start=0) +
scale_fill_manual("Gender of students with\nnet negative self-efficacies", values = c("magenta4", "green4"),
labels = c(paste("Women (", round((negSE_gender$Freq[1] / sum(negSE_gender$Freq))*100, 0), "%)", sep=""),
paste("Men (", round((negSE_gender$Freq[2] / sum(negSE_gender$Freq))*100, 0), "%)", sep=""))) +
theme_minimal() +
theme(axis.title = element_blank(), panel.border = element_blank(), panel.grid=element_blank(), axis.ticks = element_blank(), axis.text=element_blank()) +
ggtitle("Figure S4Bi") + theme(plot.title = element_text(hjust = 0.5))
ggsave(figS4bi, filename="./Figures/FigureS4Bi.svg", width = 3, height = 3)
plot(figS4bi)
Figure S4Bii
negSE_labExp = data.frame(Var1 = as.character(1:7), Freq = c(nrow(delta_net[delta_net$Value < 0 & delta_net$Exp == 1,]), nrow(delta_net[delta_net$Value < 0 & delta_net$Exp == 2,]), nrow(delta_net[delta_net$Value < 0 & delta_net$Exp == 3,]), nrow(delta_net[delta_net$Value < 0 & delta_net$Exp == 4,]), nrow(delta_net[delta_net$Value < 0 & delta_net$Exp == 5,]), nrow(delta_net[delta_net$Value < 0 & delta_net$Exp == 6,]), nrow(delta_net[delta_net$Value < 0 & delta_net$Exp == 7,])), stringsAsFactors = FALSE)
negSE_labExp$Var1 <- factor(negSE_labExp$Var1, levels = as.character(1:7))
figS4bii <- ggplot(negSE_labExp, aes(x="", y=Freq, fill = Var1))+
geom_bar(width = 1, stat = "identity") +
coord_polar("y", start=0, direction = -1) +
scale_fill_manual("Lab experience of students with\nnet negative self-efficacies", values=c("7" = "#333333", "6" = "#4C4C4C", "5" = "#666666", "4" = "#808080", "3" = "#999999", "2" = "#B2B2B2", "1" = "#CCCCCC"),
labels = c(paste("< 2 years (", round((negSE_labExp$Freq[1] / sum(negSE_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("2 - 3 years (", round((negSE_labExp$Freq[2] / sum(negSE_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("3 - 4 years (", round((negSE_labExp$Freq[3] / sum(negSE_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("4 - 5 years (", round((negSE_labExp$Freq[4] / sum(negSE_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("5 - 6 years (", round((negSE_labExp$Freq[5] / sum(negSE_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("6 - 7 years (", round((negSE_labExp$Freq[6] / sum(negSE_labExp$Freq)) * 100, 0), "%)", sep=""),
paste("> 7 years (", round((negSE_labExp$Freq[7] / sum(negSE_labExp$Freq)) * 100, 0), "%)", sep=""))) +
theme_minimal() +
theme(axis.title = element_blank(), panel.border = element_blank(), panel.grid=element_blank(), axis.ticks = element_blank(), axis.text=element_blank()) +
ggtitle("Figure S4Bii") + theme(plot.title = element_text(hjust = 0.5))
ggsave(figS4bii, filename="./Figures/FigureS4Bii.svg", width = 3, height = 3)
plot(figS4bii)
Figure S5
The only significant difference between men and women’s self-efficacy is in experimental design at the beginning of the semester in 2017.
Figure S5A
# Split data by responses to self-efficacy questionaire before and after the first semester of graduate school
SE_pre = data_2017[,c(2,14:27)] # Change 2017 / 2018
# Split by gender
femaleSE_pre = SE_pre[SE_pre$Sex == "Female",]
maleSE_pre = SE_pre[SE_pre$Sex == "Male",]
# Function to calculate the p-value of every factor on the post-test score
ordLogReg <- function(dat, Q, test) {
# Get relevant data for each question
pre = dat[,(Q+13)]
post = dat[,(Q+27)]
# Create data frame including pre-test score, post-test score, gender, and years of research experience
mat = data.frame(Pre = pre, Post = post, Gender = dat$Sex, stringsAsFactors = FALSE)
# Reorder
mat$Gender <- factor(mat$Gender, levels=c("Female", "Male"), ordered=TRUE)
if (test == "pre") {
summary(glm(Pre ~ Gender, data = mat))
} else if (test == "post") {
summary(glm(Post ~ Gender, data = mat))
} else {
cat("Please enter either 'pre' or 'post' as a selection of the test to correctly use this function.\n")
}
}
# Calculate the p-value for each question and save to an output table
Items = c("Understand contemporary concepts in your field", "Make use of the primary scientific research literature in your field (e.g., journal articles)", "Identify a specific question for investigation based on the research in your field", "Formulate a research hypothesis based on a specific question", "Design an experiment or theoretical test of the hypothesis", "Understand the importance of 'controls' in research", "Observe and collect data", "Statistically analyze data", "Interpret data by relating results to the original hypothesis", "Reformulate your original research hypothesis (as appropriate)", "Relate your results to the 'bigger picture' in your field", "Orally communicate the results of research projects", "Write a research paper for publication", "Think independently")
sigOut = data.frame(Item = Items, Question = 1:14, tValue = rep(0, 14), pValue = rep(0, 14), Significance = rep("", 14), stringsAsFactors = FALSE)
for (i in 1:14) {
ordLogRegTest = as.vector(ordLogReg(data_2017, i, "pre")$coefficients)[c(6,8)]
sigOut$tValue[i] = round(ordLogRegTest[1], 2)
sigOut$pValue[i] = round(ordLogRegTest[2], 4)
}
for (i in 1:14) {
if (sigOut$pValue[i] < 0.001) {
sigOut$Significance[i] = "***"
} else if (sigOut$pValue[i] < 0.01) {
sigOut$Significance[i] = "**"
} else if (sigOut$pValue[i] < 0.05) {
sigOut$Significance[i] = "*"
}
}
print(sigOut)## Item
## 1 Understand contemporary concepts in your field
## 2 Make use of the primary scientific research literature in your field (e.g., journal articles)
## 3 Identify a specific question for investigation based on the research in your field
## 4 Formulate a research hypothesis based on a specific question
## 5 Design an experiment or theoretical test of the hypothesis
## 6 Understand the importance of 'controls' in research
## 7 Observe and collect data
## 8 Statistically analyze data
## 9 Interpret data by relating results to the original hypothesis
## 10 Reformulate your original research hypothesis (as appropriate)
## 11 Relate your results to the 'bigger picture' in your field
## 12 Orally communicate the results of research projects
## 13 Write a research paper for publication
## 14 Think independently
## Question tValue pValue Significance
## 1 1 1.26 0.2155
## 2 2 0.75 0.4595
## 3 3 1.40 0.1674
## 4 4 0.21 0.8330
## 5 5 2.94 0.0053 **
## 6 6 1.99 0.0531
## 7 7 0.88 0.3852
## 8 8 0.18 0.8602
## 9 9 1.52 0.1363
## 10 10 1.97 0.0549
## 11 11 1.01 0.3171
## 12 12 -0.99 0.3284
## 13 13 0.98 0.3334
## 14 14 1.25 0.2190# Count number of students
numFemaleStudents = nrow(femaleSE_pre)
numMaleStudents = nrow(maleSE_pre)
# Melt responses
m_femaleSE_pre = melt(femaleSE_pre)## Using Sex as id variablesm_maleSE_pre = melt(maleSE_pre)## Using Sex as id variables# Add question number label to each array
m_femaleSE_pre$variable = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = numFemaleStudents)
m_maleSE_pre$variable = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = numMaleStudents)
# Change labels to strings for categorical plotting
m_femaleSE_pre[m_femaleSE_pre$value==1,"value"] = "a1" # Not at all, female
m_femaleSE_pre[m_femaleSE_pre$value==2,"value"] = "b1" # A little, female
m_femaleSE_pre[m_femaleSE_pre$value==3,"value"] = "c1" # A moderate amount, female
m_femaleSE_pre[m_femaleSE_pre$value==4,"value"] = "d1" # A lot, female
m_femaleSE_pre[m_femaleSE_pre$value==5,"value"] = "e1" # A great deal, female
m_maleSE_pre[m_maleSE_pre$value==1,"value"] = "a2" # Not at all, male
m_maleSE_pre[m_maleSE_pre$value==2,"value"] = "b2" # A little, male
m_maleSE_pre[m_maleSE_pre$value==3,"value"] = "c2" # A moderate amount, male
m_maleSE_pre[m_maleSE_pre$value==4,"value"] = "d2" # A lot, male
m_maleSE_pre[m_maleSE_pre$value==5,"value"] = "e2" # A great deal, male
# Combine gender dataframes for plotting
m_gSE_pre <- rbind(m_femaleSE_pre, m_maleSE_pre)
# Order variables for plotting
m_gSE_pre$variable <- factor(m_gSE_pre$variable, levels = c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), ordered = TRUE)
m_gSE_pre$Sex <- factor(m_gSE_pre$Sex , levels = c("Female", "Male"), ordered = TRUE)
m_gSE_pre$value = factor(m_gSE_pre$value, levels = c("a1", "b1", "c1", "d1", "e1", "a2", "b2", "c2", "d2", "e2"))
# Create stacked bar chart for each question, for male and female students
figS5a <- ggplot(m_gSE_pre, aes(Sex)) +
geom_bar(aes(fill=value), position = "fill") + facet_grid(~ variable) +
theme_bw() +
scale_fill_manual("Research skills\nself-efficacy", values = c("a1" = "#D5ACD9", "b1" = "#BE7DC4", "c1" = "#9A33A2", "d1" = "#67006F", "e1" = "#420047", "a2" = "#ACD9AC", "b2" = "#7DC47D", "c2" = "#33A233", "d2" = "#006F00", "e2" = "#004700"), labels = c("", "", "", "", "", "Not at all", "A little", "A moderate amount", "A lot", "A great deal")) +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
theme(axis.title=element_blank(), axis.text=element_blank(), axis.ticks=element_blank(), strip.background = element_blank()) +
guides(fill=guide_legend(ncol = 2)) +
ggtitle("Figure S5A") + theme(plot.title = element_text(hjust = 0.5))
# Plot figure
ggsave(figS5a, filename="./Figures/FigureS5A.svg", width = 6, height = 4)
plot(figS5a)
Figure S5B
# Split data by responses to self-efficacy questionaire before and after the first semester of graduate school
SE_post = data_2017[,c(2,28:41)] # Change 2017 / 2018
# Split by gender
femaleSE_post = SE_post[SE_post$Sex == "Female",]
maleSE_post = SE_post[SE_post$Sex == "Male",]
# Calculate the p-value for each question and save to an output table
Items = c("Understand contemporary concepts in your field", "Make use of the primary scientific research literature in your field (e.g., journal articles)", "Identify a specific question for investigation based on the research in your field", "Formulate a research hypothesis based on a specific question", "Design an experiment or theoretical test of the hypothesis", "Understand the importance of 'controls' in research", "Observe and collect data", "Statistically analyze data", "Interpret data by relating results to the original hypothesis", "Reformulate your original research hypothesis (as appropriate)", "Relate your results to the 'bigger picture' in your field", "Orally communicate the results of research projects", "Write a research paper for publication", "Think independently")
sigOut = data.frame(Item = Items, Question = 1:14, tValue = rep(0, 14), pValue = rep(0, 14), Significance = rep("", 14), stringsAsFactors = FALSE)
for (i in 1:14) {
ordLogRegTest = as.vector(ordLogReg(data_2017, i, "post")$coefficients)[c(6,8)]
sigOut$tValue[i] = round(ordLogRegTest[1], 2)
sigOut$pValue[i] = round(ordLogRegTest[2], 4)
}
for (i in 1:14) {
if (sigOut$pValue[i] < 0.001) {
sigOut$Significance[i] = "***"
} else if (sigOut$pValue[i] < 0.01) {
sigOut$Significance[i] = "**"
} else if (sigOut$pValue[i] < 0.05) {
sigOut$Significance[i] = "*"
}
}
print(sigOut)## Item
## 1 Understand contemporary concepts in your field
## 2 Make use of the primary scientific research literature in your field (e.g., journal articles)
## 3 Identify a specific question for investigation based on the research in your field
## 4 Formulate a research hypothesis based on a specific question
## 5 Design an experiment or theoretical test of the hypothesis
## 6 Understand the importance of 'controls' in research
## 7 Observe and collect data
## 8 Statistically analyze data
## 9 Interpret data by relating results to the original hypothesis
## 10 Reformulate your original research hypothesis (as appropriate)
## 11 Relate your results to the 'bigger picture' in your field
## 12 Orally communicate the results of research projects
## 13 Write a research paper for publication
## 14 Think independently
## Question tValue pValue Significance
## 1 1 0.16 0.8708
## 2 2 0.62 0.5356
## 3 3 1.19 0.2400
## 4 4 1.84 0.0725
## 5 5 0.72 0.4732
## 6 6 0.64 0.5243
## 7 7 0.84 0.4065
## 8 8 0.18 0.8574
## 9 9 -0.72 0.4777
## 10 10 0.20 0.8421
## 11 11 0.03 0.9740
## 12 12 -0.15 0.8844
## 13 13 0.10 0.9187
## 14 14 1.21 0.2340# Count number of students
numFemaleStudents = nrow(femaleSE_post)
numMaleStudents = nrow(maleSE_post)
# Melt responses
m_femaleSE_post = melt(femaleSE_post)## Using Sex as id variablesm_maleSE_post = melt(maleSE_post)## Using Sex as id variables# Add question number label to each array
m_femaleSE_post$variable = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = numFemaleStudents)
m_maleSE_post$variable = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = numMaleStudents)
# Change labels to strings for categorical plotting
m_femaleSE_post[m_femaleSE_post$value==1,"value"] = "a1" # Not at all, female
m_femaleSE_post[m_femaleSE_post$value==2,"value"] = "b1" # A little, female
m_femaleSE_post[m_femaleSE_post$value==3,"value"] = "c1" # A moderate amount, female
m_femaleSE_post[m_femaleSE_post$value==4,"value"] = "d1" # A lot, female
m_femaleSE_post[m_femaleSE_post$value==5,"value"] = "e1" # A great deal, female
m_maleSE_post[m_maleSE_post$value==1,"value"] = "a2" # Not at all, male
m_maleSE_post[m_maleSE_post$value==2,"value"] = "b2" # A little, male
m_maleSE_post[m_maleSE_post$value==3,"value"] = "c2" # A moderate amount, male
m_maleSE_post[m_maleSE_post$value==4,"value"] = "d2" # A lot, male
m_maleSE_post[m_maleSE_post$value==5,"value"] = "e2" # A great deal, male
# Combine gender dataframes for plotting
m_gSE_post <- rbind(m_femaleSE_post, m_maleSE_post)
# Order variables for plotting
m_gSE_post$variable <- factor(m_gSE_post$variable, levels = c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), ordered = TRUE)
m_gSE_post$Sex <- factor(m_gSE_post$Sex , levels = c("Female", "Male"), ordered = TRUE)
m_gSE_post$value = factor(m_gSE_post$value, levels = c("a1", "b1", "c1", "d1", "e1", "a2", "b2", "c2", "d2", "e2"))
# Create stacked bar chart for each question, for male and female students
figS5b <- ggplot(m_gSE_post, aes(Sex)) +
geom_bar(aes(fill=value), position = "fill") + facet_grid(~ variable) +
theme_bw() +
scale_fill_manual("Research skills\nself-efficacy", values = c("a1" = "#D5ACD9", "b1" = "#BE7DC4", "c1" = "#9A33A2", "d1" = "#67006F", "e1" = "#420047", "a2" = "#ACD9AC", "b2" = "#7DC47D", "c2" = "#33A233", "d2" = "#006F00", "e2" = "#004700"), labels = c("", "", "", "", "", "Not at all", "A little", "A moderate amount", "A lot", "A great deal"), drop = FALSE) +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
theme(axis.title=element_blank(), axis.text=element_blank(), axis.ticks=element_blank(), strip.background = element_blank()) +
guides(fill=guide_legend(ncol = 2)) +
ggtitle("Figure S5B") + theme(plot.title = element_text(hjust = 0.5))
# Plot figure
ggsave(figS5b, filename="./Figures/FigureS5B.svg", width = 6, height = 4)
plot(figS5b)
Figure S5C
# Split data by responses to self-efficacy questionaire before and after the first semester of graduate school
data_2018 = data[data$Year == 2018,]
SE_pre = data_2018[,c(2,14:27)] # Change 2017 / 2018
# Remove if gender not specified
SE_pre = SE_pre[!is.na(SE_pre$Sex),]
# Split by gender
femaleSE_pre = SE_pre[SE_pre$Sex == "Female",]
maleSE_pre = SE_pre[SE_pre$Sex == "Male",]
# Calculate the p-value for each question and save to an output table
Items = c("Understand contemporary concepts in your field", "Make use of the primary scientific research literature in your field (e.g., journal articles)", "Identify a specific question for investigation based on the research in your field", "Formulate a research hypothesis based on a specific question", "Design an experiment or theoretical test of the hypothesis", "Understand the importance of 'controls' in research", "Observe and collect data", "Statistically analyze data", "Interpret data by relating results to the original hypothesis", "Reformulate your original research hypothesis (as appropriate)", "Relate your results to the 'bigger picture' in your field", "Orally communicate the results of research projects", "Write a research paper for publication", "Think independently")
sigOut = data.frame(Item = Items, Question = 1:14, tValue = rep(0, 14), pValue = rep(0, 14), Significance = rep("", 14), stringsAsFactors = FALSE)
for (i in 1:14) {
ordLogRegTest = as.vector(ordLogReg(data_2017, i, "pre")$coefficients)[c(6,8)]
sigOut$tValue[i] = round(ordLogRegTest[1], 2)
sigOut$pValue[i] = round(ordLogRegTest[2], 4)
}
for (i in 1:14) {
if (sigOut$pValue[i] < 0.001) {
sigOut$Significance[i] = "***"
} else if (sigOut$pValue[i] < 0.01) {
sigOut$Significance[i] = "**"
} else if (sigOut$pValue[i] < 0.05) {
sigOut$Significance[i] = "*"
}
}
print(sigOut)## Item
## 1 Understand contemporary concepts in your field
## 2 Make use of the primary scientific research literature in your field (e.g., journal articles)
## 3 Identify a specific question for investigation based on the research in your field
## 4 Formulate a research hypothesis based on a specific question
## 5 Design an experiment or theoretical test of the hypothesis
## 6 Understand the importance of 'controls' in research
## 7 Observe and collect data
## 8 Statistically analyze data
## 9 Interpret data by relating results to the original hypothesis
## 10 Reformulate your original research hypothesis (as appropriate)
## 11 Relate your results to the 'bigger picture' in your field
## 12 Orally communicate the results of research projects
## 13 Write a research paper for publication
## 14 Think independently
## Question tValue pValue Significance
## 1 1 1.26 0.2155
## 2 2 0.75 0.4595
## 3 3 1.40 0.1674
## 4 4 0.21 0.8330
## 5 5 2.94 0.0053 **
## 6 6 1.99 0.0531
## 7 7 0.88 0.3852
## 8 8 0.18 0.8602
## 9 9 1.52 0.1363
## 10 10 1.97 0.0549
## 11 11 1.01 0.3171
## 12 12 -0.99 0.3284
## 13 13 0.98 0.3334
## 14 14 1.25 0.2190# Count number of students
numFemaleStudents = nrow(femaleSE_pre)
numMaleStudents = nrow(maleSE_pre)
# Melt responses
m_femaleSE_pre = melt(femaleSE_pre)## Using Sex as id variablesm_maleSE_pre = melt(maleSE_pre)## Using Sex as id variables# Add question number label to each array
m_femaleSE_pre$variable = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = numFemaleStudents)
m_maleSE_pre$variable = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = numMaleStudents)
# Change labels to strings for categorical plotting
m_femaleSE_pre[m_femaleSE_pre$value==1,"value"] = "a1" # Not at all, female
m_femaleSE_pre[m_femaleSE_pre$value==2,"value"] = "b1" # A little, female
m_femaleSE_pre[m_femaleSE_pre$value==3,"value"] = "c1" # A moderate amount, female
m_femaleSE_pre[m_femaleSE_pre$value==4,"value"] = "d1" # A lot, female
m_femaleSE_pre[m_femaleSE_pre$value==5,"value"] = "e1" # A great deal, female
m_maleSE_pre[m_maleSE_pre$value==1,"value"] = "a2" # Not at all, male
m_maleSE_pre[m_maleSE_pre$value==2,"value"] = "b2" # A little, male
m_maleSE_pre[m_maleSE_pre$value==3,"value"] = "c2" # A moderate amount, male
m_maleSE_pre[m_maleSE_pre$value==4,"value"] = "d2" # A lot, male
m_maleSE_pre[m_maleSE_pre$value==5,"value"] = "e2" # A great deal, male
# Combine gender dataframes for plotting
m_gSE_pre <- rbind(m_femaleSE_pre, m_maleSE_pre)
# Order variables for plotting
m_gSE_pre$variable <- factor(m_gSE_pre$variable, levels = c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), ordered = TRUE)
m_gSE_pre$Sex <- factor(m_gSE_pre$Sex , levels = c("Female", "Male"), ordered = TRUE)
m_gSE_pre$value = factor(m_gSE_pre$value, levels = c("a1", "b1", "c1", "d1", "e1", "a2", "b2", "c2", "d2", "e2"))
# Create stacked bar chart for each question, for male and female students
figS5c <- ggplot(m_gSE_pre, aes(Sex)) +
geom_bar(aes(fill=value), position = "fill") + facet_grid(~ variable) +
theme_bw() +
scale_fill_manual("Research skills\nself-efficacy", values = c("a1" = "#D5ACD9", "b1" = "#BE7DC4", "c1" = "#9A33A2", "d1" = "#67006F", "e1" = "#420047", "a2" = "#ACD9AC", "b2" = "#7DC47D", "c2" = "#33A233", "d2" = "#006F00", "e2" = "#004700"), labels = c("", "", "", "", "", "Not at all", "A little", "A moderate amount", "A lot", "A great deal")) +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
theme(axis.title=element_blank(), axis.text=element_blank(), axis.ticks=element_blank(), strip.background = element_blank()) +
guides(fill=guide_legend(ncol = 2)) +
ggtitle("Figure S5C") + theme(plot.title = element_text(hjust = 0.5))
# Plot figure
ggsave(figS5c, filename="./Figures/FigureS5C.svg", width = 6, height = 4)
plot(figS5c)
Figure S5D
# Split data by responses to self-efficacy questionaire before and after the first semester of graduate school
SE_post = data_2018[,c(2,28:41)] # Change 2017 / 2018
# Remove if gender not specified
SE_post = SE_post[!is.na(SE_post$Sex),]
# Split by gender
femaleSE_post = SE_post[SE_post$Sex == "Female",]
maleSE_post = SE_post[SE_post$Sex == "Male",]
# Calculate the p-value for each question and save to an output table
Items = c("Understand contemporary concepts in your field", "Make use of the primary scientific research literature in your field (e.g., journal articles)", "Identify a specific question for investigation based on the research in your field", "Formulate a research hypothesis based on a specific question", "Design an experiment or theoretical test of the hypothesis", "Understand the importance of 'controls' in research", "Observe and collect data", "Statistically analyze data", "Interpret data by relating results to the original hypothesis", "Reformulate your original research hypothesis (as appropriate)", "Relate your results to the 'bigger picture' in your field", "Orally communicate the results of research projects", "Write a research paper for publication", "Think independently")
sigOut = data.frame(Item = Items, Question = 1:14, tValue = rep(0, 14), pValue = rep(0, 14), Significance = rep("", 14), stringsAsFactors = FALSE)
for (i in 1:14) {
ordLogRegTest = as.vector(ordLogReg(data_2017, i, "post")$coefficients)[c(6,8)]
sigOut$tValue[i] = round(ordLogRegTest[1], 2)
sigOut$pValue[i] = round(ordLogRegTest[2], 4)
}
for (i in 1:14) {
if (sigOut$pValue[i] < 0.001) {
sigOut$Significance[i] = "***"
} else if (sigOut$pValue[i] < 0.01) {
sigOut$Significance[i] = "**"
} else if (sigOut$pValue[i] < 0.05) {
sigOut$Significance[i] = "*"
}
}
print(sigOut)## Item
## 1 Understand contemporary concepts in your field
## 2 Make use of the primary scientific research literature in your field (e.g., journal articles)
## 3 Identify a specific question for investigation based on the research in your field
## 4 Formulate a research hypothesis based on a specific question
## 5 Design an experiment or theoretical test of the hypothesis
## 6 Understand the importance of 'controls' in research
## 7 Observe and collect data
## 8 Statistically analyze data
## 9 Interpret data by relating results to the original hypothesis
## 10 Reformulate your original research hypothesis (as appropriate)
## 11 Relate your results to the 'bigger picture' in your field
## 12 Orally communicate the results of research projects
## 13 Write a research paper for publication
## 14 Think independently
## Question tValue pValue Significance
## 1 1 0.16 0.8708
## 2 2 0.62 0.5356
## 3 3 1.19 0.2400
## 4 4 1.84 0.0725
## 5 5 0.72 0.4732
## 6 6 0.64 0.5243
## 7 7 0.84 0.4065
## 8 8 0.18 0.8574
## 9 9 -0.72 0.4777
## 10 10 0.20 0.8421
## 11 11 0.03 0.9740
## 12 12 -0.15 0.8844
## 13 13 0.10 0.9187
## 14 14 1.21 0.2340# Count number of students
numFemaleStudents = nrow(femaleSE_post)
numMaleStudents = nrow(maleSE_post)
# Melt responses
m_femaleSE_post = melt(femaleSE_post)## Using Sex as id variablesm_maleSE_post = melt(maleSE_post)## Using Sex as id variables# Add question number label to each array
m_femaleSE_post$variable = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = numFemaleStudents)
m_maleSE_post$variable = rep(c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), each = numMaleStudents)
# Change labels to strings for categorical plotting
m_femaleSE_post[m_femaleSE_post$value==1,"value"] = "a1" # Not at all, female
m_femaleSE_post[m_femaleSE_post$value==2,"value"] = "b1" # A little, female
m_femaleSE_post[m_femaleSE_post$value==3,"value"] = "c1" # A moderate amount, female
m_femaleSE_post[m_femaleSE_post$value==4,"value"] = "d1" # A lot, female
m_femaleSE_post[m_femaleSE_post$value==5,"value"] = "e1" # A great deal, female
m_maleSE_post[m_maleSE_post$value==1,"value"] = "a2" # Not at all, male
m_maleSE_post[m_maleSE_post$value==2,"value"] = "b2" # A little, male
m_maleSE_post[m_maleSE_post$value==3,"value"] = "c2" # A moderate amount, male
m_maleSE_post[m_maleSE_post$value==4,"value"] = "d2" # A lot, male
m_maleSE_post[m_maleSE_post$value==5,"value"] = "e2" # A great deal, male
# Combine gender dataframes for plotting
m_gSE_post <- rbind(m_femaleSE_post, m_maleSE_post)
# Order variables for plotting
m_gSE_post$variable <- factor(m_gSE_post$variable, levels = c("Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11", "Q12", "Q13", "Q14"), ordered = TRUE)
m_gSE_post$Sex <- factor(m_gSE_post$Sex , levels = c("Female", "Male"), ordered = TRUE)
m_gSE_post$value = factor(m_gSE_post$value, levels = c("a1", "b1", "c1", "d1", "e1", "a2", "b2", "c2", "d2", "e2"))
# Create stacked bar chart for each question, for male and female students
figS5d <- ggplot(m_gSE_post, aes(Sex)) +
geom_bar(aes(fill=value), position = "fill") + facet_grid(~ variable) +
theme_bw() +
scale_fill_manual("Research skills\nself-efficacy", values = c("a1" = "#D5ACD9", "b1" = "#BE7DC4", "c1" = "#9A33A2", "d1" = "#67006F", "e1" = "#420047", "a2" = "#ACD9AC", "b2" = "#7DC47D", "c2" = "#33A233", "d2" = "#006F00", "e2" = "#004700"), labels = c("", "", "", "", "", "Not at all", "A little", "A moderate amount", "A lot", "A great deal"), drop = FALSE) +
theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank()) +
theme(axis.title=element_blank(), axis.text=element_blank(), axis.ticks=element_blank(), strip.background = element_blank()) +
guides(fill=guide_legend(ncol = 2)) +
ggtitle("Figure S5D") + theme(plot.title = element_text(hjust = 0.5))
# Plot figure
ggsave(figS5d, filename="./Figures/FigureS5D.svg", width = 6, height = 4)
plot(figS5d)
Figure S6
Women in the 2017 cohort were less confident in designing experiments at the beginning of the semester, despite performing comparably on an assessment of experimental design aptitude. The geneder gap in self-efficacy disappears by the end of the course.
Figure S6A
# Split data by sex, counting students
numFStudents_2017 = sum(data_2017$Sex == "Female")
numMStudents_2017 = sum(data_2017$Sex == "Male")
# Perform t-tests for pairwise combinations of tests and genders
fpre_mpre = t.test(x = (data_2017[data_2017$Sex == "Female", "Pre_CITotal"] / 14) * 100, y = (data_2017[data_2017$Sex == "Male", "Pre_CITotal"] / 14) * 100, alternative = "two.sided")
fpre_fpost = t.test(x = (data_2017[data_2017$Sex == "Female", "Pre_CITotal"] / 14) * 100, y = (data_2017[data_2017$Sex == "Female", "Post_CITotal"] / 14) * 100, alternative = "two.sided")
fpost_mpost = t.test(x = (data_2017[data_2017$Sex == "Female", "Post_CITotal"] / 14) * 100, y = (data_2017[data_2017$Sex == "Male", "Post_CITotal"] / 14) * 100, alternative = "two.sided")
mpre_mpost = t.test(x = (data_2017[data_2017$Sex == "Male", "Post_CITotal"] / 14) * 100, y = (data_2017[data_2017$Sex == "Male", "Post_CITotal"] / 14) * 100, alternative = "two.sided")
# Create variables to hold results
sigOut = data.frame(X = c("Female, Pre-test", "Female, Pre-test", "Female, Post-test", "Male, Pre-test"), Y = c("Male, Pre-test", "Female, Post-test", "Male, Post-test", "Male, Post-test"), t = rep(0, 4), pValue = rep(0, 4), Significance = rep("", 4), stringsAsFactors = FALSE)
# Enter test results into data frame
sigOut$t[1] = fpre_mpre$statistic # Female, pre-test v male, pre-test
sigOut$pValue[1] = fpre_mpre$p.value
sigOut$t[2] = fpre_fpost$statistic # Female, pre-test v female, post-test
sigOut$pValue[2] = fpre_fpost$p.value
sigOut$t[3] = fpost_mpost$statistic # Female, post-test v male, post-test
sigOut$pValue[3] = fpost_mpost$p.value
sigOut$t[4] = mpre_mpost$statistic # Male, pre-test v male, post-test
sigOut$pValue[4] = mpre_mpost$p.value
for (i in 1:4) {
if (sigOut$pValue[i] < 0.001) {
sigOut$Significance[i] = "***"
} else if (sigOut$pValue[i] < 0.01) {
sigOut$Significance[i] = "**"
} else if (sigOut$pValue[i] < 0.05) {
sigOut$Significance[i] = "*"
}
}
# Print statistical test results
print(sigOut)## X Y t pValue Significance
## 1 Female, Pre-test Male, Pre-test -0.9490637 0.34816263
## 2 Female, Pre-test Female, Post-test -1.8895807 0.06646876
## 3 Female, Post-test Male, Post-test 0.2065868 0.83731411
## 4 Male, Pre-test Male, Post-test 0.0000000 1.00000000# Calculate means and standard deviations for pre- and post-test by sex
gSE = data.frame(Test = rep(c("Pre", "Post"), 2), Gender = rep(c("Female", "Male"), each = 2), Mean = rep(0, 4), SD = rep(0, 4), stringsAsFactors = FALSE)
gSE$Mean[1] = mean((data_2017[data_2017$Sex == "Female", "Pre_CITotal"] / 14) * 100)
gSE$SD[1] = sd((data_2017[data_2017$Sex == "Female", "Pre_CITotal"] / 14) * 100)
gSE$Mean[2] = mean((data_2017[data_2017$Sex == "Female", "Post_CITotal"] / 14) * 100)
gSE$SD[2] = sd((data_2017[data_2017$Sex == "Female", "Post_CITotal"] / 14) * 100)
gSE$Mean[3] = mean((data_2017[data_2017$Sex == "Male", "Pre_CITotal"] / 14) * 100)
gSE$SD[3] = sd((data_2017[data_2017$Sex == "Male", "Pre_CITotal"] / 14) * 100)
gSE$Mean[4] = mean((data_2017[data_2017$Sex == "Male", "Post_CITotal"] / 14) * 100)
gSE$SD[4] = sd((data_2017[data_2017$Sex == "Male", "Post_CITotal"] / 14) * 100)
# Order variables for plotting
gSE$Test <- factor(gSE$Test, levels = c("Pre", "Post"), ordered = TRUE)
figS6a <- ggplot(gSE, aes(x = Test, y = Mean, fill = Gender)) +
geom_bar(position = position_dodge(), stat = "identity") +
geom_errorbar(aes(ymin = Mean - SD, ymax = Mean + SD), width = 0.3, position = position_dodge(0.9)) +
scale_fill_manual("Students' Sex", values = c("magenta4", "green4"), labels = c("Women", "Men")) +
scale_x_discrete(breaks = c("Pre", "Post"), labels = c("Pre-test", "Post-test")) +
ylab("Total BEDCI score (%)") +
theme_bw() +
theme(panel.grid.minor = element_blank(), panel.grid.major.x = element_blank(), axis.title.x = element_blank()) +
ggtitle("Figure S6A") + theme(plot.title = element_text(hjust = 0.5))
plot(figS6a)
ggsave(figS6a, file="./Figures/FigS6A.svg", width = 3, height = 5)Figure S6Bi
# Getresponses for each of the five Likert scale responses for Question 5 on the self-efficacy survey ("Design an experiment or theoretical test of the hypothesis") on pre-test
fQ5_pre = (as.vector(table(factor(data_2017[(data_2017$Sex == "Female"),"Pre_Q5"], levels = 1:5))) / nrow(data_2017[(data_2017$Sex == "Female"),])) * 100
mQ5_pre = (as.vector(table(factor(data_2017[(data_2017$Sex == "Male"),"Pre_Q5"], levels = 1:5))) / nrow(data_2017[(data_2017$Sex == "Male"),])) * 100
Q5_pre_freq = c(fQ5_pre, mQ5_pre)
# Create data frame with frequency of responses (pre-test values)
Q5_pre = data.frame(lev = rep(1:5, 2), value = Q5_pre_freq, gender = rep(c("Female", "Male"), each = 5), col = as.character(1:10), stringsAsFactors = FALSE)
Q5_pre$col <- factor(Q5_pre$col , levels = c("1", "2", "3", "4", "5", "6", "7", "8", "9", "10"), ordered = TRUE)
Q5_pre$gender <- factor(Q5_pre$gender, levels = c("Female", "Male"), ordered = TRUE)
# Get responses for each of the five Likert scale responses for Question 5 on the self-efficacy survey ("Design an experiment or theoretical test of the hypothesis") on post-test
fQ5_post = (as.vector(table(factor(data_2017[(data_2017$Sex == "Female"),"Post_Q5"], levels = 1:5))) / nrow(data_2017[(data_2017$Sex == "Female"),])) * 100
mQ5_post = (as.vector(table(factor(data_2017[(data_2017$Sex == "Male"),"Post_Q5"], levels = 1:5))) / nrow(data_2017[(data_2017$Sex == "Male"),])) * 100
Q5_post_freq = c(fQ5_post, mQ5_post)
# Create data frame with frequency of responses (post-test values)
Q5_post = data.frame(lev = rep(1:5, 2), value = Q5_post_freq, gender = rep(c("Female", "Male"), each = 5), col = as.character(1:10), stringsAsFactors = FALSE)
Q5_post$col <- factor(Q5_post$col , levels = c("1", "2", "3", "4", "5", "6", "7", "8", "9", "10"), ordered = TRUE)
Q5_post$gender <- factor(Q5_post$gender, levels = c("Female", "Male"), ordered = TRUE)
# Histogram
figS6bi <- ggplot(Q5_pre, aes(x=lev, y=value)) +
geom_bar(aes(fill = col), stat = "identity", position = position_dodge()) +
theme_bw() +
scale_fill_manual("Research skills\nself-efficacy", values = c("1" = "#D5ACD9", "2" = "#BE7DC4", "3" = "#9A33A2", "4" = "#67006F", "5" = "#420047", "6" = "#ACD9AC", "7" = "#7DC47D", "8" = "#33A233", "9" = "#006F00", "10" = "#004700"), labels = c("", "", "", "", "", "Not at all", "A little", "A moderate amount", "A lot", "A great deal")) +
guides(fill=guide_legend(ncol = 2)) +
scale_x_continuous("Self-reported self-efficacy\nwith experimental design", breaks=c(1:5) ,limits=c(0.5,5.5), labels = c("Not at all", "A little", "A moderate amount", "A lot", "A great deal")) +
ylab("Students (%)") +
theme(panel.grid.minor = element_blank(), panel.grid.major.x = element_blank()) +
ggtitle("Figure S6Bi") + theme(plot.title = element_text(hjust = 0.5))
plot(figS6bi)
ggsave(figS6bi, file="./Figures/FigS6Bi.svg", width = 5, height = 3)Figure S6Bii
# Statistical test
ordLogRegTest = as.vector(ordLogReg(data_2017, 5, "pre")$coefficients)[c(6,8)] # t value and p value
# Significance
sig1 = "is NOT a statistically significant difference"
sig2 = ""
if (ordLogRegTest[2] < 0.001) {
sig1 = "IS a statistically significant difference"
sig2 = "***"
} else if (ordLogRegTest[2] < 0.01) {
sig1 = "IS a statistically significant difference"
sig2 = "**"
} else if (ordLogRegTest[2] < 0.05) {
sig1 = "IS a statistically significant difference"
sig2 = "*"
}
# Print output
cat(paste("By ordinal logistic regression, there ", sig1, " between male and female student responses on the self-efficacy research skills survey about experimental design before their first semester of graduate school (t = ", round(ordLogRegTest[1], 2), ", p = ", round(ordLogRegTest[2], 4), sig2, ")\n", sep=""))## By ordinal logistic regression, there IS a statistically significant difference between male and female student responses on the self-efficacy research skills survey about experimental design before their first semester of graduate school (t = 2.94, p = 0.0053**)# Stacked bar chart
figS6bii <- ggplot(Q5_pre, aes(x=gender, y=value)) +
geom_bar(aes(fill=col), stat = "identity", position = "fill") +
theme_bw() +
scale_fill_manual("legend", values = c("1" = "#D5ACD9", "2" = "#BE7DC4", "3" = "#9A33A2", "4" = "#67006F", "5" = "#420047", "6" = "#ACD9AC", "7" = "#7DC47D", "8" = "#33A233", "9" = "#006F00", "10" = "#004700")) +
theme(panel.grid.minor = element_blank(), panel.grid.major = element_blank(), axis.title.x=element_blank(), axis.text.x=element_blank(), axis.title.y=element_blank(), axis.text.y=element_blank(), legend.position="none", axis.ticks = element_blank()) +
ggtitle("Figure S6Bii") + theme(plot.title = element_text(hjust = 0.5))
plot(figS6bii)
ggsave(figS6bii, file="./Figures/FigS6Bii.svg", width = 1, height = 5)Figure S6Biii
# Histogram
figS6biii <- ggplot(Q5_post, aes(x=lev, y=value)) +
geom_bar(aes(fill = col), stat = "identity", position = position_dodge()) +
theme_bw() +
scale_fill_manual("Research skills\nself-efficacy", values = c("1" = "#D5ACD9", "2" = "#BE7DC4", "3" = "#9A33A2", "4" = "#67006F", "5" = "#420047", "6" = "#ACD9AC", "7" = "#7DC47D", "8" = "#33A233", "9" = "#006F00", "10" = "#004700"), labels = c("", "", "", "", "", "Not at all", "A little", "A moderate amount", "A lot", "A great deal")) +
guides(fill=guide_legend(ncol = 2)) +
scale_x_continuous("Self-reported self-efficacy\nwith experimental design", breaks=c(1:5) ,limits=c(0.5,5.5), labels = c("Not at all", "A little", "A moderate amount", "A lot", "A great deal")) +
ylab("Students (%)") +
theme(panel.grid.minor = element_blank(), panel.grid.major.x = element_blank()) +
ggtitle("Figure S6Biii") + theme(plot.title = element_text(hjust = 0.5))
plot(figS6biii)
ggsave(figS6biii, file="./Figures/FigS6Biii.svg", width = 5, height = 3)Figure S6Biv
# Statistical test
ordLogRegTest = as.vector(ordLogReg(data_2017, 5, "post")$coefficients)[c(6,8)] # t value and p value
# Significance
sig1 = "is NOT a statistically significant difference"
sig2 = ""
if (ordLogRegTest[2] < 0.001) {
sig1 = "IS a statistically significant difference"
sig2 = "***"
} else if (ordLogRegTest[2] < 0.01) {
sig1 = "IS a statistically significant difference"
sig2 = "**"
} else if (ordLogRegTest[2] < 0.05) {
sig1 = "IS a statistically significant difference"
sig2 = "*"
}
# Print output
cat(paste("By ordinal logistic regression, there ", sig1, " between male and female student responses on the self-efficacy research skills survey about experimental design before their first semester of graduate school (t = ", round(ordLogRegTest[1], 2), ", p = ", round(ordLogRegTest[2], 4), sig2, ")\n", sep=""))## By ordinal logistic regression, there is NOT a statistically significant difference between male and female student responses on the self-efficacy research skills survey about experimental design before their first semester of graduate school (t = 0.72, p = 0.4732)# Stacked bar chart
figS6biv <- ggplot(Q5_post, aes(x=gender, y=value)) +
geom_bar(aes(fill=col), stat = "identity", position = "fill") +
theme_bw() +
scale_fill_manual("legend", values = c("1" = "#D5ACD9", "2" = "#BE7DC4", "3" = "#9A33A2", "4" = "#67006F", "5" = "#420047", "6" = "#ACD9AC", "7" = "#7DC47D", "8" = "#33A233", "9" = "#006F00", "10" = "#004700")) +
theme(panel.grid.minor = element_blank(), panel.grid.major = element_blank(), axis.title.x=element_blank(), axis.text.x=element_blank(), axis.title.y=element_blank(), axis.text.y=element_blank(), legend.position="none", axis.ticks = element_blank()) +
ggtitle("Figure S6Biv") + theme(plot.title = element_text(hjust = 0.5))
plot(figS6biv)
ggsave(figS6biv, file="./Figures/FigS6Biv.svg", width = 1, height = 5)Table S1
Demographics of student participants in the study from 2017 and 2018.
# Gender
mat = matrix(c(data3['Women', 'Year2017'], data3['Women', 'Year2018'], data3['Men', 'Year2017'], data3['Men', 'Year2018']), nrow = 2, dimnames = list("Year" = c("2017", "2018"), "Sex" = c("Female", "Male")))
cat("GENDER\n")## GENDERprop.test(t(mat), correct=FALSE)##
## 2-sample test for equality of proportions without continuity
## correction
##
## data: t(mat)
## X-squared = 1.8713, df = 1, p-value = 0.1713
## alternative hypothesis: two.sided
## 95 percent confidence interval:
## -0.32975834 0.05732244
## sample estimates:
## prop 1 prop 2
## 0.3846154 0.5208333# Race
mat = matrix(c(data3['URM', 'Year2017'], data3['URM', 'Year2018'], data3['NonURM', 'Year2017'], data3['NonURM', 'Year2018']), nrow = 2, dimnames = list("Year" = c("2017", "2018"), "Race" = c("URM", "nonURM")))
cat("RACE\n")## RACEprop.test(t(mat), correct=FALSE)##
## 2-sample test for equality of proportions without continuity
## correction
##
## data: t(mat)
## X-squared = 0.16267, df = 1, p-value = 0.6867
## alternative hypothesis: two.sided
## 95 percent confidence interval:
## -0.3437991 0.2262256
## sample estimates:
## prop 1 prop 2
## 0.4615385 0.5203252# Previous years of research lab experience
mat = matrix(c(as.matrix(table(data_2017$LabExp)), as.matrix(table(factor(data_2018$LabExp, levels = 1:7)))), nrow = 2, dimnames = list("Year" = c("2017", "2018"), "Experience" = c("0-2", "2-3", "3-4", "4-5", "5-6", "6-7", "7+")))
cat("LAB EXPERIENCE\n") ## LAB EXPERIENCEchisq.test(mat)## Warning in chisq.test(mat): Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: mat
## X-squared = 12.59, df = 6, p-value = 0.05002# Previous degree subject
mat = matrix(c(colSums(data_2017[,3:9]), colSums(data_2018[,3:9])), nrow = 7, dimnames = list("Degree" = c("Biology", "Chemistry", "Engineering", "Math", "Physics", "CompSci", "Other"), "Year" = c("2017", "2018")))
cat("DEGREE\n") ## DEGREEchisq.test(mat)## Warning in chisq.test(mat): Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: mat
## X-squared = 6.0568, df = 6, p-value = 0.4169# Program
mat = matrix(c(data3['BSPH', 'Year2017'], data3['BIG', 'Year2017'], data3['BBS', 'Year2017'], data3['BioP', 'Year2017'], data3['ChemBio', 'Year2017'], data3['SysBio', 'Year2017'], data3['Virology', 'Year2017'], data3['BSPH', 'Year2018'], data3['BIG', 'Year2018'], data3['BBS', 'Year2018'], data3['BioP', 'Year2018'], data3['ChemBio', 'Year2018'], data3['SysBio', 'Year2018'], data3['Virology', 'Year2018']), nrow = 7, dimnames = list("Program" = c("BPH", "BIG", "BBS", "BP", "CB", "SB", "V"), "Year" = c("2017", "2018")))
cat("PROGRAM\n") ## PROGRAMchisq.test(mat)## Warning in chisq.test(mat): Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: mat
## X-squared = 13.209, df = 6, p-value = 0.03984# Experience with experimental design
cat("EXPERIENCE\n")## EXPERIENCEwilcox.test(x = data_2017$ExperimetalDeignExperience, y = data_2018$ExperimetalDeignExperience, alternative = "two.sided")##
## Wilcoxon rank sum test with continuity correction
##
## data: data_2017$ExperimetalDeignExperience and data_2018$ExperimetalDeignExperience
## W = 1277, p-value = 0.849
## alternative hypothesis: true location shift is not equal to 0# Comfort with experimental design
cat("COMFORT\n")## COMFORTwilcox.test(x = data_2017$ExperimetalDeignComfort, y = data_2018$ExperimetalDeignComfort, alternative = "two.sided")##
## Wilcoxon rank sum test with continuity correction
##
## data: data_2017$ExperimetalDeignComfort and data_2018$ExperimetalDeignComfort
## W = 1306.5, p-value = 0.9946
## alternative hypothesis: true location shift is not equal to 0# Pre-test total self-efficacy in research skills
cat("PRE SELF-EFFICACY\n")## PRE SELF-EFFICACYwilcox.test(x = melt(data_2017[,14:27])$value, y = melt(data_2018[,14:27])$value, alternative = "two.sided")## No id variables; using all as measure variables
## No id variables; using all as measure variables##
## Wilcoxon rank sum test with continuity correction
##
## data: melt(data_2017[, 14:27])$value and melt(data_2018[, 14:27])$value
## W = 249748, p-value = 0.4159
## alternative hypothesis: true location shift is not equal to 0# Post-test total self-efficacy in research skills
cat("POST SELF-EFFICACY\n")## POST SELF-EFFICACYwilcox.test(x = melt(data_2017[,28:41])$value, y = melt(data_2018[,28:41])$value, alternative = "two.sided")## No id variables; using all as measure variables
## No id variables; using all as measure variables##
## Wilcoxon rank sum test with continuity correction
##
## data: melt(data_2017[, 28:41])$value and melt(data_2018[, 28:41])$value
## W = 255055, p-value = 0.9221
## alternative hypothesis: true location shift is not equal to 0# Net change in research skills self-efficacy
cat("DELTA SELF-EFFICACY\n")## DELTA SELF-EFFICACYwilcox.test(x = (melt(data_2017[,28:41])$value - melt(data_2017[,14:27])$value), y = (melt(data_2018[,28:41])$value - melt(data_2018[,14:27])$value), alternative = "two.sided")## No id variables; using all as measure variables
## No id variables; using all as measure variables
## No id variables; using all as measure variables
## No id variables; using all as measure variables##
## Wilcoxon rank sum test with continuity correction
##
## data: (melt(data_2017[, 28:41])$value - melt(data_2017[, 14:27])$value) and (melt(data_2018[, 28:41])$value - melt(data_2018[, 14:27])$value)
## W = 262874, p-value = 0.3343
## alternative hypothesis: true location shift is not equal to 0Table S2
Students significantly improved in many aspects of research skills self-efficacy during their first semester of graduate school.
# Note: this code was already included for the creation of Figure 3B
# Calculate the p-value for each question and save to an output table
Items = c("Understand contemporary concepts in your field", "Make use of the primary scientific research literature in your field (e.g., journal articles)", "Identify a specific question for investigation based on the research in your field", "Formulate a research hypothesis based on a specific question", "Design an experiment or theoretical test of the hypothesis", "Understand the importance of 'controls' in research", "Observe and collect data", "Statistically analyze data", "Interpret data by relating results to the original hypothesis", "Reformulate your original research hypothesis (as appropriate)", "Relate your results to the 'bigger picture' in your field", "Orally communicate the results of research projects", "Write a research paper for publication", "Think independently")
sigOut = data.frame(Item = Items, Question = 1:14, V = rep(0, 14), pValue = rep(0, 14), Significance = rep("", 14), stringsAsFactors = FALSE)
for (i in 1:14) {
wTest = wilcoxonSignedRankTest(data, i)
sigOut$V[i] = wTest$statistic
sigOut$pValue[i] = round(wTest$p.value, 4)
}
wTest_Total = wilcox.test(melt(SE_pre)$value, melt(SE_post)$value, paired=TRUE, alternative = "two.sided", exact=FALSE)## Using Sex as id variables
## Using Sex as id variablestmp = data.frame(Item = "Total", Question = 15, V = wTest_Total$statistic, pValue = round(wTest_Total$p.value, 4),Significance = "", stringsAsFactors = F)
sigOut[15,] = tmp
pValues = p.adjust(sigOut$pValue, method = "fdr")
for (i in 1:15) {
sigOut$pAdj[i] = round(pValues[i], 4)
if (sigOut$pAdj[i] < 0.001) {
sigOut$Significance[i] = "***"
} else if (sigOut$pAdj[i] < 0.01) {
sigOut$Significance[i] = "**"
} else if (sigOut$pAdj[i] < 0.05) {
sigOut$Significance[i] = "*"
}
}
# Print signficance table
print(sigOut) ## Item
## 1 Understand contemporary concepts in your field
## 2 Make use of the primary scientific research literature in your field (e.g., journal articles)
## 3 Identify a specific question for investigation based on the research in your field
## 4 Formulate a research hypothesis based on a specific question
## 5 Design an experiment or theoretical test of the hypothesis
## 6 Understand the importance of 'controls' in research
## 7 Observe and collect data
## 8 Statistically analyze data
## 9 Interpret data by relating results to the original hypothesis
## 10 Reformulate your original research hypothesis (as appropriate)
## 11 Relate your results to the 'bigger picture' in your field
## 12 Orally communicate the results of research projects
## 13 Write a research paper for publication
## 14 Think independently
## V Total
## Question V pValue Significance pAdj
## 1 1 524.5 0.0071 * 0.0118
## 2 2 488.0 0.0784 0.0905
## 3 3 677.0 0.0164 * 0.0224
## 4 4 429.5 0.0000 *** 0.0000
## 5 5 339.5 0.0000 *** 0.0000
## 6 6 389.0 0.0033 ** 0.0062
## 7 7 607.5 0.3072 0.3072
## 8 8 498.0 0.0004 ** 0.0012
## 9 9 385.5 0.0028 ** 0.0060
## 10 10 286.0 0.0000 *** 0.0000
## 11 11 506.0 0.1198 0.1284
## 12 12 413.0 0.0024 ** 0.0060
## 13 13 411.0 0.0208 * 0.0260
## 14 14 619.0 0.0114 * 0.0171
## V 15 31223.5 0.0000 *** 0.0000Table S3
Students significantly improved in concept inventory questions relating to controls, hypotheses, biological variation, and accuracy.
# Note: this code was already included for the creation of Figure 2C
# Calculate the p-value for each question and save to an output table
coreConcepts = c("Controls", "", "Hypotheses", "", "Biological variation", "", "Accuracy", "Extraneous factors", "", "Independent sampling", "", "Random sampling", "", "Purpose of experiments")
questions = c(1, 5, 2, 9, 3, 10, 4, 6, 14, 7, 12, 8, 13, 11) # Questions ordered by subject
sigOut = data.frame(CoreConcept = coreConcepts, Question = questions, Chi = rep(0, 14), pValue = rep(0, 14), Significance = rep("", 14), stringsAsFactors = FALSE)
i = 1 # Iterate over every row of output
for (q in questions) {
mnTest = mcNemarChiSquaredTest(data_2017, q)
sigOut$Chi[i] = round(mnTest$statistic, 2)
sigOut$pValue[i] = round(mnTest$p.value, 4)
i = i + 1
}
pValues = p.adjust(sigOut$pValue, method = "fdr")
for (i in 1:14) {
sigOut$pAdj[i] = round(pValues[i], 4)
if (sigOut$pAdj[i] < 0.001) {
sigOut$Significance[i] = "***"
} else if (sigOut$pAdj[i] < 0.01) {
sigOut$Significance[i] = "**"
} else if (sigOut$pAdj[i] < 0.05) {
sigOut$Significance[i] = "*"
}
}
# Print signficance table
print(sigOut) # Also Table S3## CoreConcept Question Chi pValue Significance pAdj
## 1 Controls 1 4.00 0.0455 0.1592
## 2 5 0.10 0.7518 1.0000
## 3 Hypotheses 2 5.88 0.0153 0.1592
## 4 9 0.00 1.0000 1.0000
## 5 Biological variation 3 0.00 1.0000 1.0000
## 6 10 4.00 0.0455 0.1592
## 7 Accuracy 4 4.92 0.0265 0.1592
## 8 Extraneous factors 6 0.12 0.7237 1.0000
## 9 14 0.12 0.7237 1.0000
## 10 Independent sampling 7 0.00 1.0000 1.0000
## 11 12 0.75 0.3865 1.0000
## 12 Random sampling 8 0.00 1.0000 1.0000
## 13 13 0.00 1.0000 1.0000
## 14 Purpose of experiments 11 0.08 0.7728 1.0000Table S4
Backgrounds of women and men participants in the study.
# Split data by sex
data_women = data[data$Sex == "Female",]
data_women = data_women[!is.na(data_women$Sex),]
data_men = data[data$Sex == "Male",]
data_men = data_men[!is.na(data_men$Sex),]
# Race
mat = matrix(c(data4['URM', 'Women'], data4['URM', 'Men'], data4['NonURM', 'Women'], data4['NonURM', 'Men']), nrow = 2, dimnames = list("Sex" = c("Women", "Men"), "Race" = c("URM", "nonURM")))
cat("RACE\n")## RACEprop.test(t(mat), correct=FALSE)##
## 2-sample test for equality of proportions without continuity
## correction
##
## data: t(mat)
## X-squared = 0.00016644, df = 1, p-value = 0.9897
## alternative hypothesis: two.sided
## 95 percent confidence interval:
## -0.2868378 0.2830854
## sample estimates:
## prop 1 prop 2
## 0.4615385 0.4634146# Previous years of research lab experience
mat = matrix(c(as.matrix(table(data_women$LabExp)), as.matrix(table(factor(data_women$LabExp, levels = 1:7)))), nrow = 2, dimnames = list("Sex" = c("Women", "Men"), "Experience" = c("0-2", "2-3", "3-4", "4-5", "5-6", "6-7", "7+")))
cat("LAB EXPERIENCE\n") ## LAB EXPERIENCEchisq.test(mat)## Warning in chisq.test(mat): Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: mat
## X-squared = 14.11, df = 6, p-value = 0.02844# Previous degree subject
mat = matrix(c(colSums(data_women[,3:9]), colSums(data_men[,3:9])), nrow = 7, dimnames = list("Degree" = c("Biology", "Chemistry", "Engineering", "Math", "Physics", "CompSci", "Other"), "Sex" = c("Women", "Men")))
cat("DEGREE\n") ## DEGREEchisq.test(mat)## Warning in chisq.test(mat): Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: mat
## X-squared = 12.209, df = 6, p-value = 0.05746# Program
mat = matrix(c(data4['BSPH', 'Women'], data4['BIG', 'Women'], data4['BBS', 'Women'], data4['BioP', 'Women'], data4['ChemBio', 'Women'], data4['SysBio', 'Women'], data4['Virology', 'Women'], data4['BSPH', 'Men'], data4['BIG', 'Men'], data4['BBS', 'Men'], data4['BioP', 'Men'], data4['ChemBio', 'Men'], data4['SysBio', 'Men'], data4['Virology', 'Men']), nrow = 7, dimnames = list("Program" = c("BPH", "BIG", "BBS", "BP", "CB", "SB", "V"), "Sex" = c("Women", "Men")))
cat("PROGRAM\n") ## PROGRAMchisq.test(mat)## Warning in chisq.test(mat): Chi-squared approximation may be incorrect##
## Pearson's Chi-squared test
##
## data: mat
## X-squared = 83.836, df = 6, p-value = 5.751e-16# Experience with experimental design
cat("EXPERIENCE\n")## EXPERIENCEwilcox.test(x = data_women$ExperimetalDeignExperience, y = data_men$ExperimetalDeignExperience, alternative = "two.sided")##
## Wilcoxon rank sum test with continuity correction
##
## data: data_women$ExperimetalDeignExperience and data_men$ExperimetalDeignExperience
## W = 1174.5, p-value = 0.5997
## alternative hypothesis: true location shift is not equal to 0# Comfort with experimental design
cat("COMFORT\n")## COMFORTwilcox.test(x = data_women$ExperimetalDeignComfort, y = data_men$ExperimetalDeignComfort, alternative = "two.sided")##
## Wilcoxon rank sum test with continuity correction
##
## data: data_women$ExperimetalDeignComfort and data_men$ExperimetalDeignComfort
## W = 1217, p-value = 0.8288
## alternative hypothesis: true location shift is not equal to 0# Pre-test total self-efficacy in research skills
cat("PRE SELF-EFFICACY\n")## PRE SELF-EFFICACYwilcox.test(x = melt(data_women[,14:27])$value, y = melt(data_women[,14:27])$value, alternative = "two.sided")## No id variables; using all as measure variables
## No id variables; using all as measure variables##
## Wilcoxon rank sum test with continuity correction
##
## data: melt(data_women[, 14:27])$value and melt(data_women[, 14:27])$value
## W = 264992, p-value = 1
## alternative hypothesis: true location shift is not equal to 0# Post-test total self-efficacy in research skills
cat("POST SELF-EFFICACY\n")## POST SELF-EFFICACYwilcox.test(x = melt(data_men[,28:41])$value, y = melt(data_men[,28:41])$value, alternative = "two.sided")## No id variables; using all as measure variables
## No id variables; using all as measure variables##
## Wilcoxon rank sum test with continuity correction
##
## data: melt(data_men[, 28:41])$value and melt(data_men[, 28:41])$value
## W = 225792, p-value = 1
## alternative hypothesis: true location shift is not equal to 0# Net change in research skills self-efficacy
cat("DELTA SELF-EFFICACY\n")## DELTA SELF-EFFICACYwilcox.test(x = (melt(data_women[,28:41])$value - melt(data_women[,14:27])$value), y = (melt(data_men[,28:41])$value - melt(data_men[,14:27])$value), alternative = "two.sided")## No id variables; using all as measure variables
## No id variables; using all as measure variables
## No id variables; using all as measure variables
## No id variables; using all as measure variables##
## Wilcoxon rank sum test with continuity correction
##
## data: (melt(data_women[, 28:41])$value - melt(data_women[, 14:27])$value) and (melt(data_men[, 28:41])$value - melt(data_men[, 14:27])$value)
## W = 256806, p-value = 0.08496
## alternative hypothesis: true location shift is not equal to 0